CN101872337A - Similarity detection method and system - Google Patents

Similarity detection method and system Download PDF

Info

Publication number
CN101872337A
CN101872337A CN 200910049990 CN200910049990A CN101872337A CN 101872337 A CN101872337 A CN 101872337A CN 200910049990 CN200910049990 CN 200910049990 CN 200910049990 A CN200910049990 A CN 200910049990A CN 101872337 A CN101872337 A CN 101872337A
Authority
CN
China
Prior art keywords
data
group
detects
similarity
groups
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN 200910049990
Other languages
Chinese (zh)
Inventor
王邕保
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.)
Semiconductor Manufacturing International Shanghai Corp
Original Assignee
Semiconductor Manufacturing International Shanghai Corp
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 Semiconductor Manufacturing International Shanghai Corp filed Critical Semiconductor Manufacturing International Shanghai Corp
Priority to CN 200910049990 priority Critical patent/CN101872337A/en
Publication of CN101872337A publication Critical patent/CN101872337A/en
Pending legal-status Critical Current

Links

Landscapes

  • Complex Calculations (AREA)

Abstract

The invention provides a similarity detection method and a system to improve the accuracy of similarity detection. The method comprises the following steps: obtaining two groups of detection data which satisfy logarithmic normal distribution; calculating the mean value and the variance of a logarithmic normal distribution function corresponding to each group of detection data; obtaining a confidence interval corresponding to each group of detection data according to the mean value and the variance; and determining the similarity of the two groups of detection data according to a rule of similarity.

Description

Similarity detection method and system
Technical field
The present invention relates to detection range, relate in particular to similarity detection method and system.
Background technology
Reliability is the critical nature of semiconductor devices, can weigh with the life-span of semiconductor devices.In semiconductor production, the life-span of each semiconductor devices that common need assurance is produced differs less, and promptly the life-span of semiconductor devices is similar.
Be that example is set forth the similarity detection scheme that present industry adopts with two batches of semiconductor devices below.
At first from two batches of devices, select the device of some respectively, and detect the life-span of the device of selecting; For cost consideration such as time of detecting and funds, the number of devices of selecting in each batch device generally is lower than 30;
Then the device lifetime that obtains according to detection, adopt F and t to add up and realize that similarity detects.
The shortcoming of above-mentioned detection scheme is: because F and t statistics generally only are applicable to large sample, the sample size that promptly needs is a lot, usually greater than 30, and above-mentionedly select that number of devices that sample promptly selects is lower than even well below 30, therefore adopt F and t statistics to realize that similarity detects and be difficult to guarantee the accuracy that detects.
Summary of the invention
The invention provides similarity detection method and system, to improve the accuracy that similarity detects.
The present invention proposes similarity detection method, in order to the similarity of the character of testing product, the method comprising the steps of: obtain two groups and detect data, described detection data fit lognormal distribution; Calculate lognormal distribution function average and variance that each group detects the data correspondence; Obtain the fiducial interval that each group detects the data correspondence according to described average and variance; According to the similarity rule, determine the similarity of two groups of data.
The invention allows for the similarity detection system, in order to the similarity of the character of testing product, this system comprises: data obtain the unit, are used to obtain two groups and detect data, described detection data fit lognormal distribution; The data computation unit is used for computational data and obtains lognormal distribution function average and the variance of respectively organizing detection data correspondence that the unit obtains; The interval unit that obtains, the average and the variance that are used for calculating according to described data computation unit obtain the fiducial interval that each group detects the data correspondence; The similarity determining unit is used for determining the similarity of two groups of data according to the similarity rule.
The present invention is based on the lognormal distribution function, detect data, described detection data fit lognormal distribution by obtaining two groups; Calculate lognormal distribution function average and variance that each group detects the data correspondence; Obtain the fiducial interval that each group detects the data correspondence according to described average and variance; And according to the similarity rule, determine the similarity of two groups of data, solved at small sample, for example detect data less than 30 situation under, existing scheme similarity detects inaccurate problem, has improved the accuracy that similarity detects.
Description of drawings
Fig. 1 is a similarity detection method process flow diagram in the embodiment of the invention;
Fig. 2 is a similarity detection system structural drawing in the embodiment of the invention.
Embodiment
Although below with reference to accompanying drawings the present invention is described in more detail, wherein represented the preferred embodiments of the present invention, should be appreciated that those skilled in the art can revise the present invention described here and still realize advantageous effects of the present invention.Therefore, following description is appreciated that extensively knowing for those skilled in the art, and not as limitation of the present invention.
For clear, whole features of practical embodiments are not described.In the following description, be not described in detail known function and structure, the unnecessary details because they can be the present invention and confusion.Will be understood that in the exploitation of any practical embodiments, must make a large amount of implementation details, for example, change into another embodiment by an embodiment according to relevant system or relevant commercial restriction to realize developer's specific objective.In addition, will be understood that this development may be complicated and time-consuming, but only be routine work to those skilled in the art.
In the following passage, with way of example the present invention is described more specifically with reference to accompanying drawing.According to the following describes and claims, advantages and features of the invention will be clearer.It should be noted that accompanying drawing all adopts very the form of simplifying and all uses non-ratio accurately, only in order to convenient, the purpose of the aid illustration embodiment of the invention lucidly.
Fig. 1 is a similarity detection method process flow diagram in the embodiment of the invention, and in conjunction with this figure, the method comprising the steps of:
Step 1 obtains two groups and detects data, described detection data fit lognormal distribution;
Present embodiment is in order to the similarity of the character of testing product, described product can be products such as semiconductor devices, described character can be the life-span of semiconductor devices etc., for example described detection data can be the lifetime datas that meets the semiconductor devices of lognormal distribution, for example (EM is tested in electron transfer, Electromigtation) and pressure migration test (SM, Stress Migration) etc.; Also can be the data of other character, only need this detection data fit lognormal distribution to get final product.Wherein no matter every group of number that detects data be greater than 30 or less than 30, and the scheme of present embodiment all has preferable effect.
Step 2 is calculated lognormal distribution function average and variance that each group detects the data correspondence;
Represent described lognormal distribution function average with μ, σ represents described lognormal distribution function variance, be that semiconductor devices is during the life-span then when detecting data, μ=ln (T50), σ=1/Slope, wherein T50 is the median life of semiconductor devices, Slope is drawn on distribute the just very much slope of straight line of coordinate diagram simulation of logarithm according to detecting data, this coordinate diagram transverse axis is the life-span, and the longitudinal axis is the accumulative total crash rate.
In the present embodiment two groups of lognormal distribution function averages that detect the data correspondence are designated as μ respectively 1And μ 2, variance is designated as σ respectively 1And σ 2
Step 3 obtains the fiducial interval that each group detects the data correspondence according to described average and variance;
This fiducial interval is defined out by the first threshold L and the second threshold value U, this fiducial interval be (L, U).
L = μ - t ( n - 1 , α ) σ n
U = μ + t ( n - 1 , α ) σ n
Wherein t can find by statistical form for adjusting coefficient, and n is for detecting the number of data, and α is a predetermined probability, is generally the confidence level of setting, and for example 95% etc.
In the present embodiment, two groups of fiducial intervals that detect the data correspondence are respectively (L 1, U 1) and (L 2, U 2), wherein
L 1 = μ 1 - t ( n 1 - 1 , α 1 ) σ 1 n 1
U 1 = μ 1 + t ( n 1 - 1 , α 1 ) σ 1 n 1
L 2 = μ 2 - t ( n 2 - 1 , α 2 ) σ 2 n 2
U 2 = μ 2 + t ( n 2 - 1 , α 2 ) σ 2 n 2
Step 4 according to the similarity rule, is determined the similarity of two groups of data.
Provide following several similarity rule to determine the similarity of two groups of data in the embodiment of the invention:
A) the lognormal distribution function average when first group of detection data is in the fiducial interval of second group of detection data correspondence, and when the lognormal distribution function average of second group of detection data was in the fiducial interval of first group of detection data correspondence, two groups were detected data for similar by force;
In the present embodiment, if μ 1∈ (L 2, U 2) and μ 2∈ (L 1, U 1), then detect data and be defined as similar by force two groups.
B) the lognormal distribution function average when first group of detection data is in the fiducial interval of second group of detection data correspondence, or second group of lognormal distribution function average that detects data be when being in first group of fiducial interval that detects the data correspondence, and two groups to detect data similar in being;
In the present embodiment, if μ 1∈ (L 2, U 2) or μ 2∈ (L 1, U 1), similar in then two groups of detection data being defined as.
C) the lognormal distribution function average when first group of detection data is not in the fiducial interval of second group of detection data correspondence, and second group of lognormal distribution function average that detects data is not in first group of fiducial interval that detects the data correspondence, but when the common factor of the fiducial interval of two groups of detection data correspondences was nonvoid set, two groups are detected data was weak similar.
In the present embodiment, if though
Figure B2009100499907D0000043
And
Figure B2009100499907D0000044
But (L1 is U1) with (L2, common factor U2) are nonvoid set, then detect data with two groups and are defined as weak similar.
D) when two groups of common factors that detect the fiducial interval of data correspondences were empty set, two groups were detected the data dissmilarities.
In the present embodiment, if
Figure B2009100499907D0000045
And
Figure B2009100499907D0000046
And (L1 is U1) with (L2, common factor U2) are empty set, then detect data with two groups and are defined as dissmilarity.
Present embodiment detects data, described detection data fit lognormal distribution by obtaining two groups; Calculate lognormal distribution function average and variance that each group detects the data correspondence; Obtain the fiducial interval that each group detects the data correspondence according to described average and variance; And according to the similarity rule, determine the similarity of two groups of data, solved at small sample, for example detect data less than 30 situation under, existing scheme similarity detects inaccurate problem, has improved the accuracy that similarity detects.
The embodiment of the invention has also proposed the similarity detection system, to improve the accuracy that similarity detects.
Fig. 2 is the structural representation of similarity detection system in the embodiment of the invention, and in conjunction with this figure, this system comprises:
Data obtain unit 21, are used to obtain two groups and detect data, described detection data fit lognormal distribution;
Data computation unit 22 is used for computational data and obtains lognormal distribution function average and the variance of respectively organizing detection data correspondence that unit 21 obtains;
The interval unit 23 that obtains, the average and the variance that are used for calculating according to described data computation unit 22 obtain the fiducial interval that each group detects the data correspondence;
Similarity determining unit 24 is used for determining the similarity of two groups of data according to the similarity rule.
This system is in order to the similarity of the character of testing product, and described product can be products such as semiconductor devices, and described character can be the life-span of semiconductor devices etc.The specific implementation process of this system, the concrete enforcement that for example relates to similarity rule, detection data etc. can obtain with reference to said method embodiment, repeats no more herein.
No matter detection system for example detects data greater than 30 number purpose samples at large sample in the present embodiment, still small sample for example detect data less than 30 situation under, can both effectively realize the detection of similarity of the character of product, thereby solved at small sample, for example detect data less than 30 situation under, existing scheme similarity detects inaccurate problem, has improved the accuracy that similarity detects.
Obviously, those skilled in the art can carry out various changes and modification to the present invention and not break away from the spirit and scope of the present invention.Like this, if of the present invention these are revised and modification belongs within the scope of claim of the present invention and equivalent technologies thereof, then the present invention also is intended to comprise these changes and modification interior.

Claims (16)

1. similarity detection method, the similarity in order to the character of testing product comprises:
Obtain two groups and detect data, described detection data fit lognormal distribution;
Calculate lognormal distribution function average and variance that each group detects the data correspondence;
Obtain the fiducial interval that each group detects the data correspondence according to described average and variance;
According to the similarity rule, determine the similarity of two groups of data.
2. the method for claim 1 is characterized in that, described fiducial interval is by first threshold L and second threshold value U definition, wherein
L = μ - t ( n - 1 , α ) σ n
U = μ + t ( n - 1 , α ) σ n
μ is described lognormal distribution function average, and σ is described lognormal distribution function variance, and t is for adjusting coefficient, and n is for detecting the number of data, and α is a predetermined probability.
3. the method for claim 1 is characterized in that, described similarity rule is:
When first group of lognormal distribution function average that detects data is in the fiducial interval of second group of detection data correspondence, and when the lognormal distribution function average of second group of detection data was in the fiducial interval of first group of detection data correspondence, two groups were detected data for similar by force.
4. the method for claim 1 is characterized in that, described similarity rule is:
When first group of lognormal distribution function average that detects data is in the fiducial interval of second group of detection data correspondence, or second group of lognormal distribution function average that detects data be when being in first group of fiducial interval that detects the data correspondence, and two groups to detect data similar in being.
5. the method for claim 1 is characterized in that, described similarity rule is:
When first group of lognormal distribution function average that detects data is not in the fiducial interval of second group of detection data correspondence, and second group of lognormal distribution function average that detects data is not in first group of fiducial interval that detects the data correspondence, but when the common factor of the fiducial interval of two groups of detection data correspondences was nonvoid set, two groups are detected data was weak similar.
6. the method for claim 1 is characterized in that, described similarity rule is: when two groups of common factors that detect the fiducial interval of data correspondences were empty set, two groups were detected the data dissmilarities.
7. the method for claim 1 is characterized in that, described detection data are the lifetime data of semiconductor devices.
8. the method for claim 1 is characterized in that, every group of number that detects data is less than 30.
9. similarity detection system, the similarity in order to the character of testing product comprises:
Data obtain the unit, are used to obtain two groups and detect data, described detection data fit lognormal distribution;
The data computation unit is used for computational data and obtains lognormal distribution function average and the variance of respectively organizing detection data correspondence that the unit obtains;
The interval unit that obtains, the average and the variance that are used for calculating according to described data computation unit obtain the fiducial interval that each group detects the data correspondence;
The similarity determining unit is used for determining the similarity of two groups of data according to the similarity rule.
10. system as claimed in claim 9 is characterized in that, described fiducial interval is by first threshold L and second threshold value U definition, wherein
L = μ - t ( n - 1 , α ) σ n
U = μ + t ( n - 1 , α ) σ n
μ is described lognormal distribution function average, and σ is described lognormal distribution function variance, and t is for adjusting coefficient, and n is for detecting the number of data, and α is a predetermined probability.
11. system as claimed in claim 9 is characterized in that, described similarity rule is:
When first group of lognormal distribution function average that detects data is in the fiducial interval of second group of detection data correspondence, and when the lognormal distribution function average of second group of detection data was in the fiducial interval of first group of detection data correspondence, two groups were detected data for similar by force.
12. system as claimed in claim 9 is characterized in that, described similarity rule is:
When first group of lognormal distribution function average that detects data is in the fiducial interval of second group of detection data correspondence, or second group of lognormal distribution function average that detects data be when being in first group of fiducial interval that detects the data correspondence, and two groups to detect data similar in being.
13. system as claimed in claim 9 is characterized in that, described similarity rule is:
When first group of lognormal distribution function average that detects data is not in the fiducial interval of second group of detection data correspondence, and second group of lognormal distribution function average that detects data is not in first group of fiducial interval that detects the data correspondence, but when the common factor of the fiducial interval of two groups of detection data correspondences was nonvoid set, two groups are detected data was weak similar.
14. system as claimed in claim 9 is characterized in that, described rule of similarity is: when two groups of common factors that detect the fiducial interval of data correspondences were empty set, two groups were detected the data dissmilarities.
15. system as claimed in claim 9 is characterized in that, described detection data are the lifetime data of semiconductor devices.
16. system as claimed in claim 9 is characterized in that, every group of number that detects data is less than 30.
CN 200910049990 2009-04-24 2009-04-24 Similarity detection method and system Pending CN101872337A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 200910049990 CN101872337A (en) 2009-04-24 2009-04-24 Similarity detection method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 200910049990 CN101872337A (en) 2009-04-24 2009-04-24 Similarity detection method and system

Publications (1)

Publication Number Publication Date
CN101872337A true CN101872337A (en) 2010-10-27

Family

ID=42997203

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 200910049990 Pending CN101872337A (en) 2009-04-24 2009-04-24 Similarity detection method and system

Country Status (1)

Country Link
CN (1) CN101872337A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103809020A (en) * 2014-01-17 2014-05-21 浙江大学 Interconnected network low-frequency oscillation frequency and damping estimation value joint confidence interval determination method
CN112348002A (en) * 2021-01-08 2021-02-09 成都云鼎智控科技有限公司 Data processing method for engine test
CN116982993A (en) * 2023-09-27 2023-11-03 之江实验室 Electroencephalogram signal classification method and system based on high-dimensional random matrix theory

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103809020A (en) * 2014-01-17 2014-05-21 浙江大学 Interconnected network low-frequency oscillation frequency and damping estimation value joint confidence interval determination method
CN103809020B (en) * 2014-01-17 2016-04-27 浙江大学 The defining method of interconnected network low-frequency oscillation frequency and damping estimated value simultaneous confidence intervals
CN112348002A (en) * 2021-01-08 2021-02-09 成都云鼎智控科技有限公司 Data processing method for engine test
CN112348002B (en) * 2021-01-08 2021-03-30 成都云鼎智控科技有限公司 Data processing method for engine test
CN116982993A (en) * 2023-09-27 2023-11-03 之江实验室 Electroencephalogram signal classification method and system based on high-dimensional random matrix theory
CN116982993B (en) * 2023-09-27 2024-04-02 之江实验室 Electroencephalogram signal classification method and system based on high-dimensional random matrix theory

Similar Documents

Publication Publication Date Title
CN103103974B (en) Method for detecting slope stability by monitoring displacement directivity parameters
Kim The relation among fit indexes, power, and sample size in structural equation modeling
CN100416543C (en) System and method for processing training data for a statistical application
US9037518B2 (en) Classifying unclassified samples
US10275333B2 (en) Risk analysis of codebase using static analysis and performance data
CN109241627B (en) Probability grading dynamic support method and device for automatically designing support scheme
CN101311738B (en) Reliability test analytical method
CN102541736B (en) Acceleration test method in software reliability execution process
CN103930886A (en) Computation of real-world error using META-analysis of replicates
WO2006014464A2 (en) Method for quantitative pcr data analysis system (qdas)
CN107133491B (en) Method for obtaining concentration of free DNA of fetus
CN103870675B (en) Load unit remnants whole life prediction system and method
CN103675022A (en) TDS detection method, detection device and detection system
CN106643965A (en) Method for accurately recognizing liquid level by template matching
CN104931989A (en) Method and device for detecting abnormal point in movement locus
CN112465001B (en) Classification method and device based on logistic regression
CN115048613A (en) Index homogenization conversion method and device, electronic equipment and storage medium
CN101872337A (en) Similarity detection method and system
CN106354940A (en) Landscape water quality simulation and early warning method based on water quality model uncertainty input
CN110214282A (en) The determination of the availability on the parking stall for motor vehicle
CN117669394B (en) Mountain canyon bridge long-term performance comprehensive evaluation method and system
CN1749988A (en) Methods and apparatus for managing and predicting performance of automatic classifiers
CN112529112B (en) Mineral identification method and device
CN111273126B (en) Power distribution network topology rapid sensing method
CN117269009A (en) Sample capacity early warning method, device, system, equipment and medium for flow cytometer

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C12 Rejection of a patent application after its publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20101027