CN110096834A - A kind of small sample life-span prediction method in short-term based on support vector machines - Google Patents
A kind of small sample life-span prediction method in short-term based on support vector machines Download PDFInfo
- Publication number
- CN110096834A CN110096834A CN201910396268.4A CN201910396268A CN110096834A CN 110096834 A CN110096834 A CN 110096834A CN 201910396268 A CN201910396268 A CN 201910396268A CN 110096834 A CN110096834 A CN 110096834A
- Authority
- CN
- China
- Prior art keywords
- life
- amount
- degradation
- sample
- short
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/17—Mechanical parametric or variational design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/04—Ageing analysis or optimisation against ageing
Abstract
The invention discloses a kind of small sample life-span prediction method in short-term based on support vector machines, comprising: sample is chosen, and amount of degradation is selected and the determination of failure criteria；Degraded data is obtained by life test in short-term；Degraded data is handled based on support vector machine method, obtains amount of degradation changing rule expression formula, and judges the multiple key parameter distributed models of amount of degradation；Based on the multiple key parameter distributed models of amount of degradation changing rule expression formula and Monte-Carlo Simulation, sample size expansion is carried out using inverse transformation method；The service life of exptended sample is calculated according to fixed failure criteria and amount of degradation changing rule expression formula, and judges Lifetime Distribution Model；Reliability assessment is carried out by selected Lifetime Distribution Model.The component of machine that the method for the invention is suitable for having obvious characteristic amount of degradation, only needed under a certain operating condition 30% time of life cycle test can the Accurate Prediction exemplar service life, the life appraisal for long-life, highly reliable research and development of products and finished product exemplar provides important channel.
Description
Technical field
The invention belongs to forecasting technique in life span fields, and in particular, to a kind of small sample in short-term based on support vector machines
Life-span prediction method.
Background technique
With the fast development of the high-end equipment such as China's Aeronautics and Astronautics, engineering machinery, to the service lifes of machine components,
Reliability requirement is higher and higher, and it is imperative to accelerate domestic high quality machine components research and development.Domestic machinery part R&D process
In, reliability demonstration takes a long time, while consuming a large amount of human and material resources, seriously affected the research and development of engineering goods into
Degree.Therefore, machine components durability test is carried out using traditional life cycle test and has been unable to satisfy research and development of products demand.
Acceleration service life test method is the important means for carrying out long-life highly reliable Survey of product life prediction, is widely used in army
The high-end equipment components research and development such as work, aerospace, vehicle.Due to experimental condition limitation and accelerated life test immutable zero
The basic demand of component failure form, even exemplar quantity is still under the conditions of accelerated test for long-life high reliable mechanical components
It is extremely limited, since small sample poor information easily causes life prediction error larger.
Summary of the invention
According to the above-mentioned deficiencies of the prior art, a kind of based on supporting vector the technical problem to be solved by the present invention is to propose
The small sample life-span prediction method in short-term of machine only needs 30% time of life cycle test can Accurate Prediction sample under a certain operating condition
Part service life, the life appraisal for long-life, highly reliable research and development of products and finished product exemplar provide important channel.
To achieve the above object, the present invention is realized according to following technical scheme:
A kind of small sample life-span prediction method in short-term based on support vector machines, which comprises the steps of:
Step 1: the selection of sample, amount of degradation is selected and the determination of failure criteria；
Step 2: obtaining degraded data by life test in short-term；
Step 3: handling based on support vector machine method degraded data, amount of degradation changing rule expression formula is obtained,
And judge the multiple key parameter distributed models of amount of degradation；
Step 4: being based on the multiple key parameter distributed models of amount of degradation changing rule expression formula and Monte-Carlo Simulation, adopt
Sample size expansion is carried out with inverse transformation method；
Step 5: the service life of exptended sample is calculated according to fixed failure criteria and amount of degradation changing rule expression formula,
And judge exptended sample Lifetime Distribution Model；
Step 6: carrying out reliability assessment by selected Lifetime Distribution Model.
Preferably, the life test in short-term in the step 2 can be the life test under exemplar original military service operating condition,
It can be the life test under exemplar accelerating mode, test life time is no less than the 30% of sample piece design service life.
Preferably, processing is carried out including using support vector machine method to degeneration number to degraded data in the step 3
According to carrying out curve fitting, the linearly or nonlinearly expression formula of amount of degradation matched curve is obtained, and then obtains its amount of degradation variation rule
Restrain multiple key parameters of expression formula.
Preferably, the sample size in the step 4 expands while including multiple keys of amount of degradation changing rule expression formula
Parameter distribution model.
Compared with the prior art, the invention has the following advantages:
A kind of small sample life-span prediction method in short-term based on support vector machines provided by the invention takes for machine components
The higher and higher situation of labour service life, reliability requirement, only needs 30% time of life cycle test can be accurate under a certain operating condition
The prediction exemplar service life shortens test period on the basis of degree of precision, reduces experimentation cost.Based on support vector machines
The life-span prediction method of small sample in short-term be the long-life, the life appraisal of highly reliable research and development of products and finished product exemplar provide it is important
Approach.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
Other attached drawings are obtained according to these attached drawings.
Fig. 1 is method flow schematic diagram of the invention；
Fig. 2 is selflubricating liner abrasion loss curve matching schematic diagram；
Fig. 3 is the first key parameter distributed model estimation schematic diagram；
Fig. 4 is K-S inspection result schematic diagram；
Fig. 5 is the second key parameter distributed model estimation schematic diagram；
Fig. 6 is K-S inspection result schematic diagram；
Fig. 7 is the abrasion loss change curve schematic diagram after sample size expands；
Fig. 8 is exptended sample Lifetime Distribution Model estimation schematic diagram；
Fig. 9 is K-S inspection result schematic diagram；
Figure 10 is selflubricating liner crash rate and Reliability Function schematic diagram.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.
As shown in Figure 1, a kind of small sample life-span prediction method in short-term based on support vector machines of the invention, including it is as follows
Step:
Step 1: the selection of sample, amount of degradation is selected and the determination of failure criteria；
Step 2: obtaining degraded data by life test in short-term；
Step 3: handling based on support vector machine method degraded data, amount of degradation changing rule expression formula is obtained,
And judge the multiple key parameter distributed models of amount of degradation；
Step 4: being based on the multiple key parameter distributed models of amount of degradation changing rule expression formula and Monte-Carlo Simulation, adopt
Sample size expansion is carried out with inverse transformation method；
Step 5: the service life of exptended sample is calculated according to fixed failure criteria and amount of degradation changing rule expression formula,
And judge Lifetime Distribution Model；
Step 6: carrying out reliability assessment by selected Lifetime Distribution Model.
Life test in short-term in step 2 can be the life test under exemplar original military service operating condition, is also possible to exemplar and adds
Life test under fast operating condition, test life time are no less than the 30% of sample piece design service life.
Processing is carried out including carrying out curve to degraded data using support vector machine method to degraded data in step 3
Fitting, obtains the linearly or nonlinearly expression formula of amount of degradation matched curve, and then obtain its amount of degradation changing rule expression formula
Multiple key parameters.
Sample size in step 4 expands while including multiple key parameter distributed modes of amount of degradation changing rule expression formula
Type
The specific embodiment of the method for the invention is further illustrated by taking certain selflubricating liner as an example below.
Embodiment, the prediction of selflubricating liner service life
It determines that selflubricating liner exemplar quantity is 4, determines tests exemplar at random by the way of purchase.
The considerations of by various aspects factor, determines that failure criteria is that abrasion loss of the selflubricating liner under operating condition of test reaches
0.228mm。
By selflubricating liner, life test measures the related data of pad wear amount at any time in short-term.
Degraded data is fitted using support vector machine method using Python software.
Partial code is as follows:
Import numpy as np
Import pandas as pd
Import sklearn.svm as SVR
Import matplotlib.pyplot as plt
Data=pd.read_excel (io='1.xlsx', header=None)
X=data.iloc [0:9000,0]
X1=np.array (x)
X=x1.reshape (x1.shape [0], 1)
Y_=data.iloc [0:9000,1]
Y=np.array (y_)
Linear_svr=SVR (kernel='linear')
linear_svr.fit(X,y)
……
The abrasion loss variation rule curve of each exemplar is as shown in Figure 2.
To the first key parameter --- it is quasi- that wear rate vector carries out exponential distribution, logarithm normal distribution and Weibull distribution etc.
It closes, obtains wear rate probability density function figure, as shown in Figure 3.The superiority and inferiority of wear rate fitting of distribution is judged using K-S inspection,
As shown in figure 4, simultaneously finally determining its distributed model.Similarly, to the second key parameter intercept --- running-in wear amount vector carries out
Processing, as shown in Figure 5, Figure 6.Available, wear rate obeys logarithm normal distribution, intercept Normal Distribution.
Based on Monte-Carlo Simulation and wear rate, intercept distributed model, selflubricating liner sample is carried out using inverse transformation method
Amount expands.According to wear rate and intercept distribution function curve, N (such as 100) a value is taken between 0-1, obtain N group wear rate and
The data of intercept, to obtain selflubricating liner abrasion loss and time curve.
Abrasion magnitude available N group life value when being failed by selflubricating liner, the sample as expanded, as shown in Figure 7.
The fitting that distributed model is carried out to the lifetime data after expansion, as shown in figure 8, judging to be distributed using K-S inspection
The superiority and inferiority of fitting, as shown in figure 9, obtaining the distribution of selflubricating liner service life meets logarithm normal distribution.
Selflubricating liner service life obeys logarithm normal distribution, is managed by the characteristic parameter and reliability of its distribution function
By the Q-percentile life being available from when lubrication liner reliability is 90% is 1700min, average life span 2100min.
Drawing selflubricating liner, failure rate estimation figure, Reliability Function figure are as shown in Figure 10 in use.
Specific embodiments of the present invention are described above.It is to be appreciated that the invention is not limited to above-mentioned
Particular implementation, those skilled in the art can make a variety of changes or modify within the scope of the claims, this not shadow
Ring substantive content of the invention.In the absence of conflict, the feature in embodiments herein and embodiment can any phase
Mutually combination.
Claims (4)
1. a kind of small sample life-span prediction method in short-term based on support vector machines, which comprises the steps of:
Step 1: the selection of sample, amount of degradation is selected and the determination of failure criteria；
Step 2: obtaining degraded data by life test in short-term；
Step 3: handling based on support vector machine method degraded data, amount of degradation changing rule expression formula is obtained, and sentence
The disconnected multiple key parameter distributed models of amount of degradation；
Step 4: the multiple key parameter distributed models of amount of degradation changing rule expression formula and Monte-Carlo Simulation are based on, using anti-
Converter technique carries out sample size expansion；
Step 5: calculating the service life of exptended sample according to fixed failure criteria and amount of degradation changing rule expression formula, and sentence
Disconnected exptended sample Lifetime Distribution Model；
Step 6: carrying out reliability assessment by selected Lifetime Distribution Model.
2. the small sample life-span prediction method according to claim 1 in short-term based on support vector machines, which is characterized in that institute
Stating the life test in short-term in step 2 is the life test under exemplar original military service operating condition or the service life under exemplar accelerating mode
Test, test life time are no less than the 30% of sample piece design service life.
3. the small sample life-span prediction method according to claim 1 in short-term based on support vector machines, which is characterized in that institute
It states in step 3 and processing is carried out including carrying out curve fitting using support vector machine method to degraded data to degraded data, obtain
To the linearly or nonlinearly expression formula of amount of degradation matched curve, and then obtain multiple crucial ginsengs of amount of degradation changing rule expression formula
Number.
4. the small sample life-span prediction method according to claim 1 in short-term based on support vector machines, which is characterized in that institute
State multiple key parameter distributed models that the sample size in step 4 expands while including amount of degradation changing rule expression formula.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910396268.4A CN110096834A (en) | 2019-05-14 | 2019-05-14 | A kind of small sample life-span prediction method in short-term based on support vector machines |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910396268.4A CN110096834A (en) | 2019-05-14 | 2019-05-14 | A kind of small sample life-span prediction method in short-term based on support vector machines |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110096834A true CN110096834A (en) | 2019-08-06 |
Family
ID=67447841
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910396268.4A Pending CN110096834A (en) | 2019-05-14 | 2019-05-14 | A kind of small sample life-span prediction method in short-term based on support vector machines |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110096834A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103472340A (en) * | 2013-09-26 | 2013-12-25 | 北京航空航天大学 | Crystal resonator storage life forecasting method based on least squares support vector machine |
| CN103971024A (en) * | 2014-05-26 | 2014-08-06 | 华北电力大学（保定） | Method for evaluating reliability of relaying protection systems under small sample failure data |
| CN109145382A (en) * | 2018-07-23 | 2019-01-04 | 长安大学 | A kind of appraisal procedure of carrier aircraft drive axle System in Small Sample Situation reliability |
| WO2019055329A1 (en) * | 2017-09-13 | 2019-03-21 | General Electric Company | A method of learning robust regression models from limited training data |
-
2019
- 2019-05-14 CN CN201910396268.4A patent/CN110096834A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103472340A (en) * | 2013-09-26 | 2013-12-25 | 北京航空航天大学 | Crystal resonator storage life forecasting method based on least squares support vector machine |
| CN103971024A (en) * | 2014-05-26 | 2014-08-06 | 华北电力大学（保定） | Method for evaluating reliability of relaying protection systems under small sample failure data |
| WO2019055329A1 (en) * | 2017-09-13 | 2019-03-21 | General Electric Company | A method of learning robust regression models from limited training data |
| CN109145382A (en) * | 2018-07-23 | 2019-01-04 | 长安大学 | A kind of appraisal procedure of carrier aircraft drive axle System in Small Sample Situation reliability |
Non-Patent Citations (1)
| Title |
|---|
| 李海滨: "工程机械用液压马达服役寿命预测", 《中国优秀硕士学位论文全文数据库 工程科技II辑》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106570281B (en) | Bayesian reliability evaluation method for small subsample product based on similar product information | |
| Liu et al. | Entropy-based sensor selection for condition monitoring and prognostics of aircraft engine | |
| CN103308723B (en) | Product service life rapid test method based on physical model | |
| LaCava et al. | Three‐dimensional bearing load share behaviour in the planetary stage of a wind turbine gearbox | |
| Colledani et al. | Performance evaluation of production lines with finite buffer capacity producing two different products | |
| CN102541736A (en) | Acceleration test method in software reliability execution process | |
| Liu et al. | Optimal coordinate sensor placements for estimating mean and variance components of variation sources | |
| Mettas et al. | Modeling and analysis of time-dependent stress accelerated life data | |
| CN112069732A (en) | Comprehensive evaluation method for artillery reliability identification test | |
| CN111400294B (en) | Data anomaly monitoring method, device and system | |
| CN110096834A (en) | A kind of small sample life-span prediction method in short-term based on support vector machines | |
| CN105488193B (en) | Method and device for predicting article popularity | |
| Nishino et al. | Taguchi's T‐method using median‐median line for small sample with outliers | |
| Vlcek et al. | Determination of rolling-element fatigue life from computer generated bearing tests | |
| CN107609234B (en) | Robustness analysis method and system based on uncertainty | |
| US20230126258A1 (en) | Machine learning device, method for generating learning models, and program | |
| CN115203840A (en) | Turbine disk reliability verification method based on Bayesian sequential test | |
| CN108052721A (en) | Carrier rocket Reliability Assessment method and device, storage medium, terminal | |
| Hidalgo et al. | A novel framework for relationship of manufacturing tolerance and component-level performance of journal bearings | |
| CN111222736A (en) | Ammunition storage reliability evaluation method based on mixed correlation vector machine model | |
| CN108572914B (en) | Method and device for generating log for pressure test | |
| CN113032998A (en) | Medical instrument life evaluation method and device | |
| CN116070972B (en) | Method and device for spot inspection of automobile parts, electronic equipment and storage medium | |
| Gu et al. | A Leadership Transfer Algorithm for the Raft | |
| Yun-fei et al. | Turbulence modeling validation in hypersonic flows |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190806 |
|
| RJ01 | Rejection of invention patent application after publication |