CN106529090B - A kind of aerospace electron class Reliability Assessment method - Google Patents
A kind of aerospace electron class Reliability Assessment method Download PDFInfo
- Publication number
- CN106529090B CN106529090B CN201611140081.0A CN201611140081A CN106529090B CN 106529090 B CN106529090 B CN 106529090B CN 201611140081 A CN201611140081 A CN 201611140081A CN 106529090 B CN106529090 B CN 106529090B
- Authority
- CN
- China
- Prior art keywords
- component
- product
- analysis
- electronic product
- reliability
- 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
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/30—Circuit design
- G06F30/36—Circuit design at the analogue level
- G06F30/367—Design verification, e.g. using simulation, simulation program with integrated circuit emphasis [SPICE], direct methods or relaxation methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/08—Probabilistic or stochastic CAD
Landscapes
- Engineering & Computer Science (AREA)
- Computer Hardware Design (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Microelectronics & Electronic Packaging (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Complex Calculations (AREA)
Abstract
The aerospace electron Reliability Assessment method based on failure mechanism that the invention discloses a kind of, in the analysis method combined using physical simulation with test, construct the physical phantom of electronic product, using means such as sensitivity analysis, Monte Carlo Analysis, stress analyses, on the basis of identification product key components, in conjunction with the actual condition of component, carry out Analysis of Failure Mechanism.Currently, for analyzing a kind of means of failure cause, to targetedly improve product, improving service life and the reliability of product after the Analysis of Failure Mechanism for component mainly failure occurs in electronic product test or use process；The present invention is to find the key characterization parameter for influencing weak device lifetime and reliability with the purpose of Analysis of Failure Mechanism, for the independent variable as reliability model.
Description
Technical field
The invention belongs to the fail-safe analyses of aerospace electron product and assessment technology field, and in particular to one kind is based on failure
The aerospace electron Reliability Assessment method of mechanism.
Background technique
Electronic product is widely used in space flight satellite and the rocket product, for example, power amplifier, calculating in carrier rocket
The single machines product such as machine, distributor, frequency modulation transmitter；The single machines products such as computer, circuit box, driver, temperature controller in satellite
Belong to electronic product.These electronic product majorities are made of printed circuit board (PCB), and circuit board has different type
Component composition, such as resistance, capacitor.
Reliability completes the index of ability as an indirect expression product job with task, can not directly survey in test
Amount needs just obtain by analysis to product and its test data and calculating, this just needs serviceability assessment technology.
So-called reliability assessment refer to using product development, test, production, use etc. during the data that are collected into and information estimate
With the reliability of evaluation product.
Currently, for reliability assessment, for the electronic product in carrier rocket, since the working time is shorter,
It is generally acknowledged that its failure type is random failure, the out-of-service time obeys exponential distribution, passes through the accumulation test period of statistical product
And failure number, to calculate the reliability of product.For satellite class product, it also hold that the failure type of product is random failure,
But since its operation on orbit time is longer, its reliability index requirements only is unable to satisfy with ground test data.Therefore, at present
For the electronic product in satellite mainly using the reliability estimation method for being based on Bayes (Bayes), by the reliable of product
Property intended result is as priori data, using ground test data as supplement information, with bayesian theory by two parts information
Carry out COMPREHENSIVE CALCULATING, thus obtain the reliability assessment of product as a result, with improve due to ground test data is insufficient and cause can
The problem of can not being verified by property index.
The either electronic product of carrier rocket or satellite is all made of the reliability estimation method based on statistics, recognizes
Failure type for product is random failure, and the reliability of product is calculated by statistical test time and failure number.For boat
The characteristics of its highly reliable System in Small Sample Situation, this method have certain limitation:
1) the reliability test period is long, at high cost
Currently, the projected life of satellite class product is increasingly longer, gradually promoted to 5-8, very from 3-5 before
To the longer time.The requirement of long-life increases difficulty with the service life of electrical type single machine and reliability demonstration to satellite.Due to grinding
The requirement of progress processed can not almost carry out the test of 1:1 on ground.In addition, going to verify its longevity by the test of a large amount of single machine grade
Life and reliability keep the cost of test high.
2) test data utilizes insufficient
Service test time and failure number two Test Informations assess product reliability, numerous for measuring in test
Critical performance parameters are unable to get use in reliability assessment, thereby result in test period relative deficiency, assessment result
It is horizontal to tend not to reflection product real reliability.
3) insufficient with the correlation degree of product
The test data as used by reliability assessment is test period, the failure cause and failure machine with product itself
Reason is without establishing direct association.Therefore, under conditions of reliability assessment result is unsatisfactory for index request, needle can not also be provided
To the Product Improvement Proposal of property, when failure number is more, illustrates that product needs to design and improve, when failure number is less or zero failure,
Reliability assessment value can only be improved by increasing test period, cause experimentation cost and waste of time.
Summary of the invention
In view of this, the characteristics of present invention is highly reliable for aerospace electron class product, System in Small Sample Situation, provides a kind of reliability
The aerospace electron class Reliability Assessment method that physics is combined with intelligence learning.For aerospace electron class product, including boat
It uses PCB and electrical type single machine etc., and the method combined using Reliability Physics with intelligence learning makes full use of the key of product
Performance parameter information assesses the reliability of product, to solve the problems, such as highly reliable Small scale product reliability assessment, and passes through weakness
Link recognition methods identifies the key components of product, provides support for the design improvement and reliability of product.
A kind of reliability estimation method of aerospace electron product of the invention, includes the following steps:
Step 1 identifies the weak link of aerospace electron product, obtains key components, specifically includes following step
It is rapid:
S11, electronic product physical simulation mould is built using circuit simulation analysis software according to electronic product physical structure
Type is arranged the input and output characteristic parameter of each component in electronic product, tentatively establishes the input and output of the electronic product
Relationship；
S12, judge electronic product with the presence or absence of incipient fault；Incipient fault if it exists, after being improved to electronic product
It performs the next step, if it does not exist incipient fault, directly execution S13；
Wherein, judge that electronic product whether there is incipient fault using Monte Carlo Method, method particularly includes:
1) each component is generated according to formula (1) and formula (2) according to the distribution parameter of each component normal distribution respectively
Performance random number, wherein i-th of component input and output characteristic parameter random number generated is denoted as Xi；
Xi=μ+σ Xi' (2)
Wherein, UjIt is the random number between [0,1], μ is mean value, and σ is standard deviation；
2) by the corresponding random number X of each componentiIt is updated in the physical phantom in S11, by simulation calculation, obtains
Result is exported to electronic product；
3) 1) step and 2) step n times are repeated, obtains M circuit output result, wherein M at least takes 50 times；
M obtained simulation data result is analyzed, the extreme value exported, and according to the output of electronic product
Energy index request, judges whether simulation data result meets performance indicator requirement, if extreme value in claimed range, judges to produce
Product are met the requirements, and incipient fault is not present；Conversely, there are incipient faults；
S13, using stress analysis means, identify the weak link of electronic product, as key components, specifically:
Under conditions of electronic product works normally, with the electronic product physical phantom established in step S11, meter
The working stress of each component in product is calculated, and obtains the working stress of each component and the ratio of specified value, and according to ratio
Sequence from big to small is ranked up each component；One or more component in following three conditions will be met to sentence
It is set to weak link:
Condition 1: the ratio is greater than 1 component；
Condition 2: by observing ratio, it is unsatisfactory for the component of drop volume requirement；
Condition 3: the difference of the ratio of two neighboring component is more than 30% and the forward component ratio that sorts is greater than 0.5
When, determine that two components sort forward component as weak link, while determining sequence before the weak link device
All components be weak link；
In the case where above 3 conditions are not satisfied, determine that first three component of sequence is weak link；
Step 2, for the Primary Component determined in step 1, in conjunction with its operating condition, with Analysis of Failure Mechanism method,
Determine the failure mechanism of key components；Then according to the failure mechanism of key components, the working environment of combination product, analysis
And determine key components performance parameter relevant to failure mechanism and working life, i.e. life characteristics parameter；
Step 3 carries out device level test, the life characteristics ginseng of measurement component during the test for key components
Number, and record its time to failure of component TF；Then the corresponding relationship of time to failure TF Yu life characteristics parameter are established, i.e.,
Obtain Reliable Mathematics model；
Step 4 is directed to different samples, each life characteristics parameter of component in Reliable Mathematics model is collected, for each
The sample variation of parameter calculates the distribution pattern distribution parameter corresponding with distribution pattern of each life characteristics parameter；
Step 5, the reliability assessment based on Latin Hypercube Sampling calculate, specifically:
For each life characteristics parameter, frequency in sampling N is first determined according to required precision, is then obtained according to step 4
The distribution pattern and distribution parameter of each life characteristics parameter carry out n times sampling using Latin Hypercube Sampling method, and will sampling
Sample is updated to respectively in the Reliable Mathematics model of step 3 foundation, and the calculated value of N number of time TF is calculated；By counting N
A TF is calculated, and is obtained TF probability-distribution function F (t), and according to the task time t of productmRequirement, calculate the reliability of product
R, it may be assumed that
R(tm)=P { TF > tm}=1-F (tm)。
Further, it after step S11 establishes input and the output relation of the electronic product, is tested, is surveyed by reality
The actual operating conditions and working condition of each component are measured, the input/output relation to electronic product are verified, to physical simulation mould
The input and output characteristic parameter of each component is adjusted in type, after adjustment, then executes S12.
Further, after the step S11 obtains input/output relation, based on the relationship to member each in electronic product
Device carries out sensitivity analysis, and obtaining influences most sensitive component to electronic product output, and just according to sensitivity,
Each component is ranked up；In the step 12, chooses the preceding some components of sensitivity sequence and carry out Monte Carlo
Analysis.
Further, after the step S11 obtains input/output relation, based on the relationship to member each in electronic product
Device carries out sensitivity analysis, and obtaining influences most sensitive component to electronic product output, and just according to sensitivity,
Each component is ranked up；In method of the step 12 to product improvement, according to component sensitivity ranking results, choosing
It takes sensitivity to sort forward component, is replaced or retrofits, to realize the improvement to electronic product.
Preferably, being used for the Analysis of Failure Mechanism method of electronic product: electric test, micro- shape in the step 2
Looks analysis, microstructure analysis, physical property detection, Microanalysis, stress test or dissection sample preparation analysis method.
Preferably, it is corresponding with life characteristics parameter to establish TF using the learning algorithm of neural network or support vector machines
Relationship, i.e. reliability mathematical model.
Preferably, the frequency in sampling N meets in the step 5:In formula, γ is required precision, σ2For sample
This variance is calculated by step 4；zαFor normal distribution quantile.
The invention has the following beneficial effects:
(1) worst case analysis (WCA) thinking is used for reference and efficiently used, identifies electronic product weak link, is reliability
Assessment provides support；Worst case analysis is that China's aerospace electron product (usually circuit) often carries out a circuit analysis work
Make, Sensitive Apparatus, and the worst feelings being likely to occur by Monte-Carlo Simulation come prediction circuit are identified using sensitivity analysis
Condition combination, so that identification circuit whether there is incipient fault.The present invention is effectively utilized this in weak link identification process
Analytical mathematics.But on the basis of identifying incipient fault, which is to pass through stress analysis to the main purpose of circuit analysis
The methods of identification circuit most weak device, and Analysis of Failure Mechanism is carried out to weak device, determines that the service life of weakness device is special
Parameter is levied, and constructs reliability model, to assess the reliability level of product.
(2) even closer in conjunction with the performance of electronic product based on the reliability assessment of failure mechanism, be conducive to design
It improves and is promoted with reliability；In the analysis method combined using physical simulation with test, the physical simulation of electronic product is constructed
Model, using the means such as sensitivity analysis, Monte Carlo Analysis, stress analysis, on the basis of identifying product key components, knot
The actual condition of component is closed, Analysis of Failure Mechanism is carried out.Currently, the Analysis of Failure Mechanism for component is mainly in electronics
After there is failure in product testing or use process, for analyzing a kind of means of failure cause, to targetedly improve
Product improves service life and the reliability of product.
The present invention is to find the key spy for influencing weak device lifetime and reliability with the purpose of Analysis of Failure Mechanism
Parameter is levied, for the independent variable as reliability model.In addition, after reliability assessment, if the reliability of electronic product is not up to
Reliability index requirements, then with Analysis of Failure Mechanism as a result, improved to product, to improve service life of product and reliable
Property, so that it is met reliability index requirements.
(3) test data needed for reliability assessment mostlys come from the performance data of component grade, and non-product complete machine
Test, experimentation cost are controlled；Currently, in order to verify the reliability of aerospace electron product whether index request, usually with list
Machine is that unit carries out overall test, and due to the limitation of experimentation cost and lead time, the product for participating in test is difficult to failure,
Test data used in reliability assessment is almost 0 failure.On the one hand, this kind of reliability test method be not due to accomplishing product failure
(Censoring), it is difficult to obtain product true service life and reliability, on the other hand, the entire single machine of this kind of approach application carries out examination
It tests, will increase experimentation cost.In addition, only being commented using the accumulative test period of product with failure quantity in reliability assessment
The reliability of the yield by estimation product, what test data utilized is not enough, and causes the waste of data information.
The present invention is in view of the above problems, pass through the Primary Component of identification electronic product, and emphasis is test with Primary Component
Object is the different magnitude of bias test of test variable development with Analysis of Failure Mechanism life characteristics parameter obtained, then with
A small amount of single machine test is as verifying.Using this test data as input, using intelligence learning algorithm, fitting building reliability mould
Type assesses the reliability of product, can effectively save experimentation cost, the compression test time, and component test during obtain
Each life characteristics parameter and complete lifetime data are obtained, it is more abundant to the more accurate of reliability description of product, data utilization.
(4) algorithm combined with intelligence learning with Latin Hypercube Sampling realizes the Reliability modeling of electronic product
With assessment；One important application of intelligence learning algorithm (such as neural network and support vector machines) is data fitting.It draws
Fourth hypercube sampling (LHS) is a kind of methods of sampling that can be improved simulation efficiency.
The algorithm that the present invention is combined with intelligence learning with Latin Hypercube Sampling realizes that the reliability of electronic product is built
Mould and assessment.Using intelligence learning algorithm as Reliability modeling tool, the test data of product is handled, to construct failure
Functional relation between preceding time (TF) and life characteristics parameter, and using this model as foundation, joined by counting each life characteristics
Several distribution characteristics with LHS thought, and is applied in reliability assessment, using sampling simulation algorithm assessment product
Reliability.
In conclusion integrated use of the present invention circuit simulation analysis, Analysis of Failure Mechanism, intelligent algorithm, Latin
The technical methods such as hypercube sampling assess the reliability of product, and can be effective to identify the weak link of aerospace electron product
Electronic product reliability experimentation cost and period are reduced, provides technical support for the design improvement and reliability of product.
Detailed description of the invention
Fig. 1 is the reliability estimation method implementation flow chart that Reliability Physics of the invention is combined with intelligence learning；
Fig. 2 is the implementation flow chart of the invention based on the identification of virtual test weak link；
Fig. 3 is electronic product Analysis of Failure Mechanism process of the invention；
Fig. 4 is the signal modulation module circuit course of work of certain type safety governor in the embodiment of the present invention；
Fig. 5 is the signal modulation module circuit physical simulation model of certain type safety governor in the embodiment of the present invention, wherein
(a) (b) (c) is three parts split；
Fig. 6 is key point signal waveform detail view in the embodiment of the present invention；
Fig. 7 is 50 Monte-Carlo Simulation result statistical results charts of DPSKCA signal virtual value in the embodiment of the present invention；
Fig. 8 is neural network structure figure in the embodiment of the present invention；
Fig. 9 is neural network fitting result chart in the embodiment of the present invention；
Figure 10 is probability-distribution function curve graph in the embodiment of the present invention.
Specific embodiment
The present invention will now be described in detail with reference to the accompanying drawings and examples.
The invention proposes the reliability estimation methods that a kind of Reliability Physics is combined with intelligence learning, high for space flight
The reliability assessment of reliable small sample electronic product, implementing procedure are as shown in Figure 1.
Reliability assessment the following steps are included:
Step 1, the weak link identification based on virtual test
The physical model that electronic product to be analyzed is constructed using virtual test method, the input and output for describing electronic product are closed
System, and verified using reality, physical model is improved to the accuracy of product description.On this basis, utilization is sensitive
The means such as degree analysis, Monte-Carlo Simulation and stress analysis, identify the weak link of product, that is, restrict the electronic product service life
Key components.The implementing procedure of weak link identification based on virtual test is as shown in Figure 2.
S11, physical phantom building, establish circuit input and output relation
According to electronic product physical structure (circuit diagram), software is analyzed using circuit simulation, builds physical phantom, if
The input and output characteristic parameter for setting each component tentatively establishes input and the output relation of product, executes step S12.
S12, simulation parameter amendment, improve simulation result accuracy
It is tested by reality, measures the actual operating conditions and working condition of each component, verified to the defeated of electronic product
Enter output relation.When deviation occurs in physical phantom, need for real test data, to each in physical phantom
The input and output characteristic parameter of component is adjusted, to improve the standard of the showed product input/output relation of physical phantom
True property.
After the completion of simulation parameter amendment, S13 is executed.
S13, sensitivity analysis identify sensitive device, provide information for Monte-Carlo Simulation
This concept of sensitivity refers to the variation of sensor output as one of the index of statement Sensor
It measures and the ratio between the input variable quantity for causing the variable quantity.It is frequently used for the static characteristic of description sensor, what it was characterized is output
Measure the reflection degree to input quantity.
For electronic product, output of products is influenced, other than inputting parameter, the input and output feature of each component
Parameter can also have a certain impact to the output of product.The sensitivity analysis of electronic product mainly reflects component input and output
Characteristic parameter changes the analysis to electronic product output performance Index Influence degree (including size and Orientation) to be analyzed.Pass through spirit
Basis of sensitivity analysis, sensitivity of the confirmation properties of product to each composition component input and output characteristic parameter variation.Pass through sensitivity point
Analysis, which can be obtained, influences most sensitive component to electronic product output, and according to sensitivity height, to each component
It is ranked up.
The specific implementation method of sensitivity analysis is detailed in GJB/Z 223 " worst-case circuit analysis guide ".
After the completion of sensitivity analysis, S14 is executed.
S14, Monte-Carlo Simulation analysis, calculate the wave of electronic product output parameter under the influence of each component is uncertain
Dynamic range judges product with the presence or absence of incipient fault accordingly
Due to production technology, material etc., input and output of the same model even with a batch of different components are special
Having a certain difference property of parameter is levied, shows certain uncertainty, according to historical experience, which is usually expressed as
Normal distribution.
It is special using covering using normal distribution as distribution pattern according to the uncertainty of each component input and output characteristic parameter
Caro emulates sampling algorithm, realizes propagation of each device performance unascertained information in electronic product, to describe each first device
Influence of the fluctuation of part input and output characteristic parameter to electronic product output characteristics, obtains the fluctuation model of electronic product output parameter
It encloses.
In order to improve Monte-Carlo Simulation efficiency, usually chooses the M component that sensitivity is sorted forward and carries out discretization,
Other lower components of sensitivity, then take the mean value of its input-output characteristic parameter.
The implementation method of electronic product Monte-Carlo Simulation analysis are as follows:
1) according to the distribution parameter (mean μ, standard deviation sigma) of component normal distribution each in circuit, according to formula (1) and public affairs
Formula (2) generates the performance random number of each component respectively.Each component input and output characteristic parameter random number generated is denoted as
(X1,X2,...,XM)。
Xi=μ+σ Xi' (2)
Wherein, UjIt is the random number between [0,1].
2) by (X1,X2,...,XM) be brought into physical phantom, by simulation calculation, obtain electronic product output knot
Fruit.
3) 1) step and 2) step n times are repeated, N number of circuit output result is obtained.
Under normal circumstances, simulation times N is bigger, and the effect of Monte-Carlo Simulation is better, can more reflect that product whether there is
Incipient fault.But simulation times are more, and it is longer to analyze the consumed time.It is the characteristics of according to space flight Small scale product, general to emulate
Number should be no less than 50 times, and the output bias for the electronic product that N number of normal assembly is completed is simulated with this.
After Monte-Carlo step emulation, obtained N number of simulation data result is analyzed, the extreme value exported
(maximum value max, minimum value min etc.), and according to the output performance index request of electronic product, whether judge simulation data result
Meet performance indicator requirement.If extreme value in claimed range, judges that product is met the requirements, incipient fault is not present, executes
S16；If extreme value has exceeded the range of requirement, the product is judged there are incipient fault, needs further to analyze, and executes S15.
S15, support is provided as a result, improving for product design using sensitivity analysis
If learning electronic product there are incipient fault by Monte-Carlo Simulation analysis, needs to be designed product and change
Into.
Sensitivity reflects the influence sensitivity that the variation of component input and output characteristic parameter inputs electronic product,
It, can be more significant to the improvement of output of products if carrying out the type or performance boost to higher sensitivity component.Therefore, in order to
Product improvement efficiency is improved, sensitivity analysis is can refer to and sorts forward component as a result, choosing sensitivity, be preferentially replaced
Or remodeling, to improve the output characteristics of electronic product.
After the completion of product improvement, S16 is executed.
S16, stress analysis identify the weak link of electronic product
Under conditions of electronic product works normally, with electronic product physical phantom, each first device in product is calculated
The actual stress of part, and judge whether the actual working stress of each component has been more than rated value and whether has met according to rated value
The requirement of drop volume.Working stress includes the working stress of limit and the working stress of transient condition.
Calculate the working stress of each component and the ratio of specified value, and according to sequence from big to small to each component into
Row sequence, ranking results are the output of stress analysis.The forward component that sorts is the weak link of product.Weak link
Really fixed condition includes:
Condition 1: the ratio is greater than 1 component；
Condition 2: by observing ratio, it is unsatisfactory for the component of drop volume requirement；(national army wherein, is asked for an interview to about drop volume
With the regulation of " component derating criteria " in standard)
Condition 3: the difference of the ratio of two neighboring component is more than 30% and the forward component ratio that sorts is greater than 0.5
When, determine that two components sort forward component as weak link, while determining sequence before the weak link device
All components be weak link；
In the case where being unsatisfactory for last condition, determine that the component of sequence preceding 3 is weak link.
Step 2, Analysis of Failure Mechanism and determination
For key components, the mistake of key components is determined with Analysis of Failure Mechanism method in conjunction with its operating condition
Imitate mechanism.
Electric test, microstructure analysis, microstructure are specifically included that for the Analysis of Failure Mechanism method of electronic product
Analysis, physical property detection, Microanalysis, stress test, dissection sample preparation.Analysis appropriate can be chosen according to products characteristics
Method determines component failure mechanism.The Analysis of Failure Mechanism principle of component is first to carry out non-destructive analysis, after carry out brokenly
The analysis of bad property；First external analysis, rear internal (dissection) analysis；First investigate the situation related with failure (route, stress item
Part, failure phenomenon etc.), post analysis failure component.According to the analysis method of failure mechanism, binding analysis principle determines failure
The analysis process of mechanism is as shown in Figure 3.According to the failure mechanism of key components, the working environment of combination product is analyzed and true
Fixed performance parameter relevant to failure mechanism and working life, i.e. life characteristics parameter.
Step 3, the Reliable Mathematics model construction based on intelligence learning
Carry out targeted device level test for key components, the life characteristics parameter during measurement test,
And record its time to failure (TF).Under normal circumstances, the life characteristics parameter of TF and component has non-linear relation, for
This case establishes the pass corresponding with life characteristics parameter TF using the intelligence learnings algorithm such as neural network, support vector machine
System, i.e. reliability mathematical model.
By taking multilayer neural network method as an example, reliability mathematical model construction method, but not limited to this method.Multilayer
The implementing procedure of neural network are as follows:
1) variable-definition
Input layer number, hidden layer number, the output layer neuron number of neural network are defined, and is determined initial
Weight ωni、ωij。
2) forward-propagating of working signal
According to hidden layer transmission function and output layer transmission function, it is input with each parameter of input layer, calculates output, and count
Calculate network error.
3) backpropagation of error signal
Using neural network BP training algorithm, weight is successively reversely adjusted along network.Firstly, adjustment hidden layer and output layer
Weight ωij, then, error signal propagated forward, to the weight ω between input layer and hidden layerniIt is adjusted.
2) and 3) 4) repeat, until error is met the requirements.
By the intelligence learnings algorithm such as neural network, the mapping relations of TF Yu life characteristics parameter, i.e. reliability number are established
Model is learned, using the life characteristics parameter of component as input, TF is as output.
Step 4, test data statistical analysis
For different samples, each life characteristics parameter of component in Reliable Mathematics model is collected, for each parameter
Sample variation calculates the mean value and variance, point for calculating each performance parameter of each life characteristics parameter using Statistical Inference
Cloth type and distribution parameter etc..
The many factors such as each life characteristics parameter and its processing technology and materials variances due to component are related, according to
Each general Normal Distribution of life characteristics parameter of experience component, also needs to be fitted data inspection.Disobeying normal state
When distribution, statistical inference need to be carried out according to data, to determine distribution pattern distribution parameter corresponding with distribution pattern.
Step 5, the reliability assessment based on Latin Hypercube Sampling calculate
It can in conjunction with space product height according to Reliable Mathematics model for the probability-distribution function of each life characteristics parameter
By the characteristics of, use for reference Latin Hypercube Sampling (Latin hypercube sampling, LHS) thought, using based on LHS's
Reliability estimation method, to improve the efficiency of Reliability evaluation.
S51, sampling dimension is determined
The component life characteristics number of parameters M for including according to Reliable Mathematics model determines sampling dimension M.
S52, hypercube is established
A hypercube is established, if hypercube is as follows: the dimension of variable is M,I=1,2 ...,
M；Wherein, xiFor i-th dimension variable,WithThe respectively upper and lower boundary of i-th dimension variable usually defines
S53, frequency in sampling is determined
It is required according to precision γ, determines frequency in sampling N.It is required thatIn formula, σ2For sample variance, step can be passed through
Rapid 4 are calculated；zαFor normal distribution quantile, can be obtained by inquiry GB4086.1 " normal distribution of statistical distribution numerical tabular "
It arrives；1- α is confidence level, and related to required precision, α generally takes 0.01~0.1.
S54, region division
It will be per one-dimensional variable xiDomain sectionIt is divided into N number of equal minizone, it may be assumed thatAn original hypercube is thus divided into M × N number of small
Hypercube.
S55, matrix is defined
The matrix A of a M × N is generated, each column of A are a random fully intermeshing of array { 1,2 ..., N }.
S56, random number is generated
Every row of A just corresponds to a selected small hypercube, random to produce in each selected small hypercube
A raw sample, thus selects N sample.
S57, reliability is calculated
According to the distribution pattern of life characteristics parameter each in Reliable Mathematics model, N number of sample is brought at random respectively
In number calculation formula (calculation formula of common distribution random numbers is as shown in table 1), the random number of each life characteristics parameter is calculated, and
The random number generated every time is updated in corresponding function, the calculated result of TF is calculated.
The random number calculation method of 1 typical probability of table distribution
Note: Ui is the i-th column of matrix A, and other parameters are the distribution parameter of corresponding distribution
It repeats the above sample calculation process and just obtains the calculated result of N number of TF after n times are sampled.It is N number of by counting
TF is calculated, and is obtained TF probability-distribution function F (t), and according to the task time t of productmRequirement, calculate the reliability R of product,
That is:
R(tm)=P { TF > tm}=1-F (tm) (3)
Embodiment:
It is illustrated by taking the signal modulation module circuit of space flight type safety governor as an example.Modulator block is spaceborne distant
The component part of measurement equipment, major function are to complete the modulation and switching output of dpsk signal.When the task of the modulator block
Between tmIt is required that be 12 years, reliability requirement 0.99.
Step 1, the weak link identification based on virtual test
S11, physical phantom building
Signal modulation module carries out shaping to the CFA (differential code) and 2ftA (twice of subcarrier signal) of input first, then
It send to phase shift frequency dividing circuit, real-time dpsk signal (DPSKCA) is generated after filtering and modulation, the course of work is as shown in Figure 4.
Physical phantom is established, as shown in Figure 5 using circuit analysis software according to circuit diagram.In artificial circuit
In model, CFA is difference code signal, and 2ftA is twice of subcarrier signal, and Reshape is the waveform by shaping, Phase_
Shift is the waveform that phase shift is generated by monostable flipflop, and Frequency_division is the waveform by frequency divider,
Filtering is the waveform after filtering, and DPSKCA is by the modulated real-time dpsk signal of A machine.
S12, simulation parameter amendment
By the actual measurement to circuit board, simulation parameter is corrected, and static work is carried out by simulation software
Make point analysis, verifies the correctness of simulation model.In emulation, the selection transient working time is 1ms.After emulation, obtain
The normal working voltage waveform detail view of circuit is as shown in Figure 6.It will be apparent from this figure that at CF code jump, sine wave
The reversed 180 degree of phase；CF code is continuous sine wave, meets the feature of dpsk signal without jump.
S13, sensitivity analysis
The main object of the sensitivity analysis of the circuit board is the DPSKCA signal voltage virtual value of output by component
Parameter perturbation variation influence, such as resistance 3R4A change in resistance influence.The results are shown in Table 2 for sensitivity analysis.From with
Upper sensitivity analysis result, which can be seen that, influences more sensitive device for DPSKCA signal virtual value size for resistance
3R11A, 3R8A, 3R9A, 3R7A, capacitor 3C2A, 3C3A.V2VAR and VAR2V is data transformation interface, is not paid attention to herein.
Table 2, sensitivity analysis result
The sensitivity of Fig. 4 DPSKCA signal virtual value
S14, Monte-Carlo Simulation analysis
By upper section sensitivity analysis as a result, Monte Carlo Analysis for signal modulation module circuit, answers root first
According to the initial deviation value of each component input and output characteristic parameter, the initial deviation of device is set, and initial deviation obeys normal state
Distribution, the distribution parameter and distribution pattern of each component are as shown in table 3.
Table 3, sensitive component input and output characteristic parameter table
It is 50 times that simulation times, which are arranged, as shown in Figure 7 using monte carlo simulation methodology calculated result.
The simulation result practiced shooting by 50 times can be seen that resistance 3R11A, 3R8A, 3R9A, 3R7A, capacitor 3C2A, 3C3A
Input and output characteristic parameter when changing in device initial deviation, the virtual value of the DPSKCA signal of output is in 0.7V-1.6V
Interior variation meets product requirement, therefore, judges that product is met the requirements, and incipient fault is not present.
S16, stress analysis
Stress data according to each component is as shown in table 4.
Table 4, component stress data table
The stress value of the above device is updated in emulation analysis parameter, the device obtained under the nominal operating condition of circuit is practical
Stress intensity, as shown in table 5.
Table 5, device actual stress size
It can be seen that circuit without overstress device from above-mentioned stress analysis result, but the real work electric current of 54HC14
The critical state for having reached the 50mA of device maximum operating currenbt, according to GJB/Z35-93 " component derating criteria ", 54HC14
Though non-overstress, stress ratio 100% have been more than the drop volume of MOS type digital circuit derating criteria level-one as defined in GJB/Z35-93
80% output electric current derating level requirement, for the Primary Component of the circuit board.
Step 2, Analysis of Failure Mechanism and determination
54HC14 is a kind of high-speed cmos type digital integration chip.Analysis of Failure Mechanism is first according to the chip in signal tune
Operating condition in circuit processed carries out simulation test.After the chip failure, using Analysis of Failure Mechanism method, the device is determined
The failure cause and mechanism of part.By technical methods such as electric test, microstructure analysis, with automatic test equipment (ATE),
The equipment such as scanning electron microscope (SEM), analyzing its failure mechanism is mainly the threshold value that hot carrier injection effect causes device
The drift of the parameters such as voltage, until failure.Life characteristics parameter relevant to the failure mechanism specifically includes that work by analysis
Two parameters of temperature T and supply current Icc.
Step 3, the Reliable Mathematics model construction based on intelligence learning
For 54HC14 chip, carry out component test, chooses different operating temperature and supply current and carry out test, receive
Collect correlation test data, comprising: supply current Icc when time to failure TF, work temperature and work, partial data such as table 6
It is shown.With multilayer neural network method, the mapping relations of T, Icc and TF of chip are constructed.
According to the life characteristics parameter and test data of 54HC14 chip, determines that input layer number is 2, export node layer
Number is 1.This example is calculated using BP neural network algorithm.According to input layer number, output layer number of nodes, consider simultaneously
The Generalization Capability of neural network fitting determines that the number of hidden nodes is 20.Neural network structure is as shown in Figure 8.
By neural network the Fitting Calculation, Reliable Mathematics model is obtained, fitting effect is as shown in Figure 9.It can be with from figure
Find out, fitting correlation coefficient 0.9646, fitting effect is preferable.
Step 4, test data statistical analysis
It is analyzed and is monitored by the circuit board different to muti-piece, the work temperature of statistics 54HC14 chip in the circuit board
Spend supply current Icc when T and work.Gaussian fitting test, two inspections are carried out to data using the Jarque-Bera method of inspection
Testing result h value is 0, shows the two equal Normal Distributions of life characteristics parameter.Parameter Estimation is carried out using statistic algorithm, is determined
43.2 DEG C of mean value of work temperature are 6.12 with standard deviation；It is 3.16 that supply current Icc mean value 45.5mA, which is with standard deviation,.
Step 5, the reliability assessment based on Latin Hypercube Sampling calculate
According to the statistic analysis result of life characteristics parameter, in conjunction with the Reliable Mathematics model of component, using based on drawing
The reliability of the reliability assessment calculation method evaluation circuit board of fourth hypercube sampling.
According to the number of life characteristics parameter, determine that sampling dimension is 2, and establishing hypercube accordingly is [x1,x2].Root
According to 0.99 reliability requirement, the reliability assessment required precision γ of the circuit board is 0.01, and taking α is 0.05, according to sampling
Frequency in sampling needed for the calculation method of number calculates is 7530, i.e. frequency in sampling should be at least 7530, just can guarantee 0.01
Computational accuracy.For this case, the frequency in sampling for determining that this is calculated is chosen for 10000.
Random number is generated using Latin Hypercube Sampling method, and is updated in Reliable Mathematics model respectively, is calculated
To 10000 TF values.It is calculated by counting N number of TF, it is as shown in Figure 10 to obtain TF probability distribution graph, and calculate according to formula (3)
Reliability assessment result is 0.996.
In conclusion the present embodiment is reliable using the aerospace electron class product that Reliability Physics is combined with intelligence learning
Property appraisal procedure to the signal modulation module circuit of space flight type safety governor carried out fail-safe analysis and assessment, by point
Analysis is concluded that
1) it is analyzed by sensitivity analysis and Monte-Carlo Simulation it is found that the product does not deposit incipient fault；
2) by stress analysis, determine that the Primary Component of signal modulation module circuit is 54HC14 chip；
3) pass through Analysis of Failure Mechanism it is found that causing the dominant mechanism of 54HC14 chip failure for hot carrier in jection effect
It answers, fails closely related with operating temperature and supply current；
It 4) is 0.996 by the reliability that signal modulation module circuit is calculated in Reliability Evaluation Algorithm.
In conclusion the above is merely preferred embodiments of the present invention, being not intended to limit the scope of the present invention.
All within the spirits and principles of the present invention, any modification, equivalent replacement, improvement and so on should be included in of the invention
Within protection scope.
Claims (7)
1. a kind of reliability estimation method of aerospace electron product, which comprises the steps of:
Step 1 identifies the weak link of aerospace electron product, obtains key components, specifically comprises the following steps:
S11, electronic product physical phantom is built using circuit simulation analysis software according to electronic product physical structure, if
The input and output characteristic parameter for setting each component in electronic product tentatively establishes input and the output relation of the electronic product；
S12, judge electronic product with the presence or absence of incipient fault；Incipient fault if it exists executes after improving to electronic product
In next step, incipient fault if it does not exist, directly execution S13；
Wherein, judge that electronic product whether there is incipient fault using Monte Carlo Method, method particularly includes:
1) property of each component is generated respectively according to formula (1) and formula (2) according to the distribution parameter of each component normal distribution
Energy random number, wherein i-th of component input and output characteristic parameter random number generated is denoted as Xi；
Xi=μ+σ X 'i (2)
Wherein, UjIt is the random number between [0,1], μ is mean value, and σ is standard deviation；
2) by the corresponding random number X of each componentiIt is updated in the physical phantom in S11, by simulation calculation, obtains electricity
Sub- output of products result；
3) 1) step and 2) step n times are repeated, obtains M circuit output result, wherein M at least takes 50 times；
M obtained simulation data result is analyzed, the extreme value exported, and is referred to according to the output performance of electronic product
Mark requires, and judges whether simulation data result meets performance indicator requirement, if extreme value in claimed range, judges that product is full
Foot requires, and incipient fault is not present；Conversely, there are incipient faults；
S13, using stress analysis means, identify the weak link of electronic product, as key components, specifically:
Under conditions of electronic product works normally, with the electronic product physical phantom established in step S11, calculates and produce
The working stress of each component in product, and obtain the working stress of each component and the ratio of specified value, and according to ratio from big
Each component is ranked up to small sequence；One or more component in following three conditions will be met to be determined as
Weak link:
Condition 1: the ratio is greater than 1 component；
Condition 2: by observing ratio, it is unsatisfactory for the component of drop volume requirement；
Condition 3: the difference of the ratio of two neighboring component is more than 30% and sequence forward component ratio when being greater than 0.5, really
Fixed two components sort forward component as weak link, while determining that sequence is all before the weak link device
Component is weak link；
In the case where above 3 conditions are not satisfied, determine that first three component of sequence is weak link；
Step 2, for the Primary Component determined in step 1, in conjunction with its operating condition, with Analysis of Failure Mechanism method, determine
The failure mechanism of key components；Then according to the failure mechanism of key components, the working environment of combination product is analyzed and true
Determine key components performance parameter relevant to failure mechanism and working life, i.e. life characteristics parameter；
Step 3 carries out device level test for key components, measures the life characteristics parameter of component during the test,
And record its time to failure of component TF；Then establish the corresponding relationship of time to failure TF and life characteristics parameter to get
To Reliable Mathematics model；
Step 4 is directed to different samples, each life characteristics parameter of component in Reliable Mathematics model is collected, for each parameter
Sample variation, calculate the distribution pattern distribution parameter corresponding with distribution pattern of each life characteristics parameter；
Step 5, the reliability assessment based on Latin Hypercube Sampling calculate, specifically:
For each life characteristics parameter, frequency in sampling N is first determined according to required precision, each longevity then obtained according to step 4
The distribution pattern and distribution parameter for ordering characteristic parameter carry out n times sampling using Latin Hypercube Sampling method, and by sampling samples
It is updated in the Reliable Mathematics model of step 3 foundation respectively, the calculated value of N number of time TF is calculated；By counting N number of TF
It calculates, obtains TF probability-distribution function F (t), and according to the task time t of productmRequirement, calculate the reliability R of product,
That is:
R(tm)=P { TF > tm}=1-F (tm)。
2. a kind of reliability estimation method of aerospace electron product as described in claim 1, which is characterized in that the step 1
In, after step S11 establishes input and the output relation of the electronic product, is tested by reality, measure the reality of each component
Border operating condition and working condition verify the input/output relation to electronic product, to component each in physical phantom
Input and output characteristic parameter be adjusted, after adjustment, then execute S12.
3. a kind of reliability estimation method of aerospace electron product as described in claim 1, which is characterized in that in the step
After S11 obtains input/output relation, sensitivity analysis is carried out to component each in electronic product based on the relationship, is obtained to electricity
Sub- output of products influences most sensitive component, and according to sensitivity height, is ranked up to each component；In the step
In rapid 12, chooses the preceding some components of sensitivity sequence and carry out Monte Carlo analysis.
4. a kind of reliability estimation method of aerospace electron product as described in claim 1, which is characterized in that in the step
After S11 obtains input/output relation, sensitivity analysis is carried out to component each in electronic product based on the relationship, is obtained to electricity
Sub- output of products influences most sensitive component, and according to sensitivity height, is ranked up to each component；In the step
In the method for rapid 12 pairs of product improvements, according to component sensitivity ranking results, chooses sensitivity and sort forward component, into
Row replacement or remodeling, to realize the improvement to electronic product.
5. a kind of reliability estimation method of aerospace electron product as described in claim 1, which is characterized in that the step 2
In, it is used for the Analysis of Failure Mechanism method of electronic product: electric test, microstructure analysis, microstructure analysis, physics
Performance detection, Microanalysis, stress test or dissection sample preparation analysis method.
6. a kind of reliability estimation method of aerospace electron product as described in claim 1, which is characterized in that the step 3
In, using the learning algorithm of neural network or support vector machines, establish the corresponding relationship of TF Yu life characteristics parameter, i.e. reliability
Mathematical model.
7. a kind of reliability estimation method of aerospace electron product as described in claim 1, which is characterized in that the step 5
In, the frequency in sampling N meets:In formula, γ is required precision, σ2For sample variance, calculated by step 4
It arrives；zαFor normal distribution quantile.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611140081.0A CN106529090B (en) | 2016-12-12 | 2016-12-12 | A kind of aerospace electron class Reliability Assessment method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611140081.0A CN106529090B (en) | 2016-12-12 | 2016-12-12 | A kind of aerospace electron class Reliability Assessment method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106529090A CN106529090A (en) | 2017-03-22 |
| CN106529090B true CN106529090B (en) | 2019-06-14 |
Family
ID=58341952
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611140081.0A Expired - Fee Related CN106529090B (en) | 2016-12-12 | 2016-12-12 | A kind of aerospace electron class Reliability Assessment method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106529090B (en) |
Families Citing this family (32)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107169165B (en) * | 2017-04-14 | 2020-07-07 | 北京航空航天大学 | Surface modification process reliability evaluation method based on environmental effect |
| CN107239594B (en) * | 2017-04-26 | 2020-11-27 | 中车株洲电力机车研究所有限公司 | Analog circuit dispersibility optimization method based on PSPICE |
| CN106991252A (en) * | 2017-04-27 | 2017-07-28 | 上海理工大学 | Sophisticated testing Evaluation of Uncertainty method |
| CN107621786A (en) * | 2017-08-16 | 2018-01-23 | 中国电子科技集团公司第十八研究所 | A kind of space power system control device worst case analysis method |
| CN108776294B (en) * | 2018-06-01 | 2020-02-14 | 北京航空航天大学 | Circuit board service life evaluation method based on self-adaptive strategy |
| CN108804813B (en) * | 2018-06-07 | 2022-07-26 | 上海空间推进研究所 | Reliability assessment method for space rail-controlled engine |
| CN108984925B (en) * | 2018-07-25 | 2023-04-18 | 北京航空航天大学 | Parallel computing method for integrated simulation analysis of reliability of electronic product |
| CN109271660B (en) * | 2018-07-31 | 2023-08-11 | 上海空间推进研究所 | Reliability evaluation method for movable connection structure of rocket engine spray pipe |
| CN109325287B (en) * | 2018-09-17 | 2023-02-07 | 中国人民解放军海军工程大学 | Method for estimating reliability parameters of mechanical part |
| CN109492282A (en) * | 2018-10-29 | 2019-03-19 | 北京遥感设备研究所 | A kind of DC/DC power module life assessment Primary Component determines method |
| CN109633467A (en) * | 2018-12-18 | 2019-04-16 | 上海精密计量测试研究所 | A kind of aerospace lithium battery managing chip reliability verification method |
| CN109766613B (en) * | 2018-12-29 | 2020-10-27 | 西安交通大学 | Probability theory-based method for predicting bullet jump of thin-wall structure under thermal noise load |
| CN110321631B (en) * | 2019-07-02 | 2023-04-07 | 江苏科技大学 | Method for evaluating machining reliability of quality characteristics of marine diesel engine body |
| CN110298126B (en) * | 2019-07-04 | 2021-11-02 | 北京航空航天大学 | Multi-element Copula power device reliability evaluation method based on failure physics |
| CN110705139A (en) * | 2019-08-27 | 2020-01-17 | 华东光电集成器件研究所 | Electronic equipment reliability evaluation method based on multi-stress coupling |
| CN110795351B (en) * | 2019-10-29 | 2023-02-28 | 中国科学院微小卫星创新研究院 | Reliability increase testing and evaluating method for component-based star software |
| CN111222204B (en) * | 2019-11-12 | 2022-10-28 | 中国航天标准化研究所 | Joint simulation design method for performance and reliability of aerospace machinery and electrical product |
| CN111060795B (en) * | 2019-11-22 | 2022-03-04 | 中国空间技术研究院 | Method for evaluating extreme low-temperature characteristics of CMOS (complementary Metal oxide semiconductor) device |
| CN113094863A (en) * | 2019-12-23 | 2021-07-09 | 南京航空航天大学 | Civil aircraft system operation reliability assessment method considering failure propagation |
| CN111553062B (en) * | 2020-04-17 | 2023-06-27 | 中国南方电网有限责任公司超高压输电公司检修试验中心 | Method and system for judging whether insulator ball head is separated |
| CN112101432B (en) * | 2020-09-04 | 2022-06-07 | 西北工业大学 | Material microscopic image and performance bidirectional prediction method based on deep learning |
| CN112541321B (en) * | 2020-12-18 | 2024-05-31 | 中国空间技术研究院 | Method and device for early screening and risk prediction of aerospace seal integrated circuit |
| CN112684324A (en) * | 2020-12-30 | 2021-04-20 | 无锡市同步电子科技有限公司 | Method for rapidly exciting and verifying faults of PCB for airborne electronic controller |
| CN112699494B (en) * | 2021-01-08 | 2024-05-28 | 北京空间飞行器总体设计部 | Reliability prediction method under maintenance support of manned spacecraft |
| CN113779694A (en) * | 2021-08-23 | 2021-12-10 | 武汉理工大学 | Manufacturing process reliability modeling method and device based on support vector machine |
| CN113468039B (en) * | 2021-08-30 | 2021-12-17 | 深圳荣耀智能机器有限公司 | Reliability evaluation method and related equipment |
| CN114139482A (en) * | 2021-09-06 | 2022-03-04 | 苏州宽温电子科技有限公司 | EDA circuit failure analysis method based on depth measurement learning |
| CN114329108B (en) * | 2022-03-15 | 2022-06-28 | 武汉力通通信有限公司 | Production condition analysis method and system based on chip functional circuit simulation test |
| CN114355094B (en) * | 2022-03-18 | 2022-06-10 | 中国电子产品可靠性与环境试验研究所（（工业和信息化部电子第五研究所）（中国赛宝实验室）） | Product reliability weak link comprehensive evaluation method and device based on multi-source information |
| CN114580118B (en) * | 2022-03-25 | 2023-12-15 | 兰州空间技术物理研究所 | Quantitative evaluation method for service life and reliability of ion thruster |
| CN115577542B (en) * | 2022-10-17 | 2023-11-10 | 中国航发沈阳发动机研究所 | Model data driven aviation complex structure and reliability fusion design method |
| CN117873760B (en) * | 2024-03-12 | 2024-05-10 | 苏州元脑智能科技有限公司 | Failure rate assessment method, device and system and readable storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104317990A (en) * | 2014-10-09 | 2015-01-28 | 中国运载火箭技术研究院 | Multi-stage task spacecraft reliability improving method based on risks |
| CN104462700A (en) * | 2014-12-15 | 2015-03-25 | 中国航空综合技术研究所 | Electronic product reliability simulation test method based on physics of failure |
| KR20160045292A (en) * | 2014-10-17 | 2016-04-27 | 동우 화인켐 주식회사 | Complex film having high heat resistance, fabrication method thereof and use of the same |
| CN106021769A (en) * | 2016-05-30 | 2016-10-12 | 北京航空航天大学 | Typical aerospace component PID establishment method |
-
2016
- 2016-12-12 CN CN201611140081.0A patent/CN106529090B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104317990A (en) * | 2014-10-09 | 2015-01-28 | 中国运载火箭技术研究院 | Multi-stage task spacecraft reliability improving method based on risks |
| KR20160045292A (en) * | 2014-10-17 | 2016-04-27 | 동우 화인켐 주식회사 | Complex film having high heat resistance, fabrication method thereof and use of the same |
| CN104462700A (en) * | 2014-12-15 | 2015-03-25 | 中国航空综合技术研究所 | Electronic product reliability simulation test method based on physics of failure |
| CN106021769A (en) * | 2016-05-30 | 2016-10-12 | 北京航空航天大学 | Typical aerospace component PID establishment method |
Non-Patent Citations (1)
| Title |
|---|
| 航天电子设备可靠性评估方法研究;孙鹏 等;《空间科学学报》;20121231(第32期);265-269 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106529090A (en) | 2017-03-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106529090B (en) | A kind of aerospace electron class Reliability Assessment method | |
| US20200327435A1 (en) | Systems and methods for sequential power system model parameter estimation | |
| CN112382582B (en) | Wafer test classification method and system | |
| Rahi et al. | Power system voltage stability assessment through artificial neural network | |
| CN103245881A (en) | Power distribution network fault analyzing method and device based on tidal current distribution characteristics | |
| CN103020459B (en) | A kind of cognitive method of various dimensions electricity consumption behavior and system | |
| CN105572572B (en) | Analog-circuit fault diagnosis method based on WKNN-LSSVM | |
| CN112052172B (en) | Rapid test method and device for third-party channel and electronic equipment | |
| CN105786678A (en) | Relevance model-based testability prediction method | |
| CN105138770B (en) | Space product Reliablility simulation appraisal procedure based on indirect characteristic quantities | |
| CN110059714A (en) | Diagnosis Method of Transformer Faults based on multi-category support vector machines | |
| CN113484813B (en) | Intelligent ammeter fault rate prediction method and system under multi-environment stress | |
| CN110174878A (en) | A kind of unmanned intelligence equipment is healthy and ensures integrated management general-purpose system | |
| CN110874685A (en) | Intelligent electric energy meter running state distinguishing method and system based on neural network | |
| Yuan et al. | A Bayesian approach to degradation-based burn-in optimization for display products exhibiting two-phase degradation patterns | |
| Frank et al. | Extracting operating modes from building electrical load data | |
| Lin et al. | A novel product remaining useful life prediction approach considering fault effects | |
| Biswas et al. | Reducing test execution cost of integrated, heterogeneous systems using continuous test data | |
| US20070180411A1 (en) | Method and apparatus for comparing semiconductor-related technical systems characterized by statistical data | |
| Ma et al. | Measurement error assessment for smart electricity meters under extreme natural environmental stresses | |
| CN112926686B (en) | BRB and LSTM model-based power consumption anomaly detection method and device for big power data | |
| CN103076556B (en) | Method for selecting function-maintenance testing points of avionic assembly | |
| Jiao et al. | An on-line anomaly identifying method for calibration devices in an automatic verification system for electricity smart meters | |
| CN105741184A (en) | Transformer state evaluation method and apparatus | |
| CN115728588B (en) | Electromagnetic compatibility detection system and method based on big data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | 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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190614 Termination date: 20211212 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |