CN107862348A - A kind of collection of illustrative plates similarity calculating method based on white wine characteristic - Google Patents
A kind of collection of illustrative plates similarity calculating method based on white wine characteristic Download PDFInfo
- Publication number
- CN107862348A CN107862348A CN201711277895.3A CN201711277895A CN107862348A CN 107862348 A CN107862348 A CN 107862348A CN 201711277895 A CN201711277895 A CN 201711277895A CN 107862348 A CN107862348 A CN 107862348A
- Authority
- CN
- China
- Prior art keywords
- mtd
- spectrum data
- white wine
- sample
- mrow
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
A kind of collection of illustrative plates similarity calculating method based on white wine characteristic, by normalize pre-process eliminate Wine Sample spectrum data scale differ greatly caused by weight difference, the spectrum data big to white wine characteristic contribution rate is filtered out by principal component analysis again, weights when determining that it participates in Similarity Measure according to the contribution rate size of spectrum data.When calculating the similarity of same type of two kinds of white wine collection of illustrative plates, the spectrum data filtered out need to be only selected, corresponding weights, which are brought into, can try to achieve Similarity value in similarity formula.The Similarity value obtained using the method for the invention can eliminate that contribution rate is low or incoherent spectrum data is to the adverse effect of Similarity value, fully take into account difference of the different characteristic composition to the contribution rate of white wine characteristic, the similarity degree of white wine characteristic can be more accurately embodied, it is more accurate to the objective evaluation of white wine.
Description
Technical field
The present invention relates to technical field of image processing, more particularly to a kind of detection method of monitor video anomalous event.
Background technology
White wine raw material, bent class, distiller's yeast and environment used in brewing process are not quite similar, even if with batch of wine
Larger difference may be produced on flavor formation.At present, China has determined that the fragrant liquor to be formed up to more than ten is planted, such as
Luzhou-flavor, Maotai-flavor, delicate fragrance type, rice-fragrant type, sesame-flavor, medicine odor type, medicinal-flavor, special odor type etc..White wine is a variety of chemistry
The mixture of composition, in addition to 98% water and ethanol, in addition to：Alcohols, esters, acids, amino acids, ethers, aldehydes etc. are micro-
Measure composition.Micro constitutent accounts for 2%, but species is a lot, with the progress of science and technology, it has now been found that decide the fragrance of white wine
The micro constitutent of style about more than 300 is planted, and is quantitatively analyzed out wherein having more than 180 to plant.
Micro constitutent is to form the material base of white wine characteristic feature, and content of the micro constitutent in wine body and proportioning determine
The odor type of white wine, mouthfeel and quality.The subjective assessment of white wine quality is mainly taster with the sense such as vision, smell, sense of taste
Feel that organ is observed the color of white wine, analyzed, describing to make overall merit.Because sense organ resolving accuracy is limited, behaviour
Make that specification is inconsistent, the operation factors such as artificial so that white wine quality is difficult to differentiate between, and grade is difficult to define.Objective evaluation is mainly right
The finger-print of white wine realizes analysis, judgement and identification to white wine etc. with the method for Mathematical treatment.The finger-print of white wine
Refer to obtain the collection of illustrative plates or image that can represent white wine characteristic by technological means such as chromatogram or spectrum, including gas chromatography,
The methods of liquid chromatography, gas chromatography mass spectrometry, infra-red sepectrometry, electronic nose.White wine collection of illustrative plates can be considered as one and is characterized as by white wine
The characteristic of white wine can be showed, preferably reacted by the n-dimensional vector of element composition, the similarity of collection of illustrative plates with objective numeral
The similarity degree of two finger-prints.Similarity calculating method common at present includes correlation coefficient process, Cosin method, Euclidean
Furthest Neighbor etc..These similarity calculating methods look at all compositions in collection of illustrative plates of equal importance, the weights such as all the components ginseng
With similarity is calculated.However, different micro constitutents are different to the fragrance contribution of white wine, different flavor, different wine storage time
Certain difference in the composition of micro constitutent be present with the white wine of different cultivars.For example, ethyl hexanoate, ethyl butyrate and lactic acid
The Esters such as ethyl ester play a leading role in liquor flavor composition, determine the principal character of white wine, alcohols and aldehyde material point
Not Wei helping for white wine fragrant composition and put fragrant composition, they are relatively low to white wine signature contributions.The main body fragrant component of aromatic Chinese spirit is
Ethyl hexanoate, the main body fragrant component of fen-flavor type white spirit is ethyl acetate.The similarity of white wine is calculated using the above method,
Determine that principal character composition and other compositions (contribution rate low or incoherent composition) equal strength of white wine quality is treated, will certainly
Influence the numerical value of similarity.On the other hand, different principal character compositions is also different to the contribution of white wine quality.Therefore,
Difference in terms of the similarity that the weights such as all the components participate in being calculated can not effectively be reflected into white wine characteristic.
The content of the invention
It is an object of the invention to solve the defects of above-mentioned prior art is present, there is provided a kind of more accurate white wine collection of illustrative plates
Similarity calculating method.
For achieving the above object, the technical solution adopted in the present invention is：
A kind of collection of illustrative plates similarity calculating method based on white wine characteristic, comprises the following steps：
(1) n different samples of similar white wine are chosen, choose p shared spectrum datas respectively to n sample；The figure
Modal data refers to the content data of white wine composition；
(2) the p spectrum data is normalized, obtains the figure after the p normalization by the n sample
The data matrix of modal data composition
(3) principal component analysis is carried out to the spectrum data matrix X, filters out the h figure big to white wine characteristic contribution rate
Modal data, h are integer and 1≤h≤P；
(4) data matrix formed to the h spectrum data by n sample carries out principal component analysis, obtains the h
The weights of each spectrum data in individual spectrum data；
(5) similarity of two Wine Samples is obtained by h spectrum data of screening and corresponding weights.
Preferably, the step (2) comprises the following steps：
A. the p shared spectrum datas are normalized, obtain the spectrum data vector x of each samplei=
(xi,1,xi,2,xi,3,…,xi,p), wherein 1≤i≤n, element xi,pFor the spectrum data after i-th of sample, p-th of normalization；
The above-mentioned spectrum data vector composition data matrix of b.n sample
Wherein, xn,pAfter p-th of normalization for representing n-th of white wine sample
Spectrum data.
Preferably, the step (3) comprises the following steps：
C. principal component analysis is carried out to the data matrix X, obtains matrix X covariance matrixWherein element
xk,i、xk,jIt is i-th and j spectrum data of k-th of sample respectively,It is the average value of i-th and j spectrum data of n sample respectively, 1≤k≤n, 1≤i≤P, 1≤j≤P, obtains square
Battle array S eigenvalue λ and by its descending arrangement, given threshold T, finds eigenvalue λm, make λm>=T, λm+1＜ T, λmIt is corresponding
Orthogonalization unit character vector am=(am,1,…,am,g,…,am,p), λmCorresponding principal component FmPass through the table of white wine spectrum data
It is F up to formulam=am,1x1+…+am,gxg+…+am,pxp, wherein am,1,…,am,g,…,am,pFor FmCharacteristic vector coefficient；
D. above-mentioned principal component F is foundmThe characteristic vector coefficient a of middle maximum absolute valuem,gCorresponding spectrum data xg, delete n
The spectrum data x of each sample in individual white wine sampleg, then each p-1 spectrum data of white wine sample residual；
E. n sample of remaining p-1 spectrum data is repeated the above steps, deletes the spy of a maximum absolute value every time
Spectrum data corresponding to vectorial coefficient is levied, untill the default h spectrum data of each sample residual.
Preferably, the step (4) comprises the following steps：
F. the spectrum data vector y of each white wine sample is formed with h spectrum data of screeningi=(yi,1,yi,2,
yi,3,…,yi,h), 1≤i≤n, data matrix Y is formed with the above-mentioned spectrum data vector of n sample；
G. principal component analysis is carried out to above-mentioned matrix Y, k principal component before taking, makes its contribution rate of accumulative total G (k) >=85%, G
(k-1) ＜ 85%, the contribution rate α of each principal component in k principal component is calculatedi, obtain by k principal component and contain h collection of illustrative plates number
According to collection of illustrative plates relational matrixWherein, the contribution rate of accumulative total G (k) byCalculate, the contribution rate αiByCalculate；
H. the vectorial b=(b of the contribution margin composition of the h spectrum data screened1,…,bt,…bh), b in formulatFor t-th
The contribution margin of spectrum data, 1≤t≤h, bt=| α1a1,t|+|α2a2,t|+…+|αkak,t|；
I. vectorial b is normalized to obtain normalized contribution rate coefficient vector c=(c1,…,ct,…ch), 1≤t≤
H, element c in formulat=bt/ sum (b), it is that rate coefficient, namely t during Similarity Measure are contributed in the normalization of t-th of spectrum data
The weights of individual spectrum data.
Preferably, the step (5) comprises the following steps：
J. the two h spectrum datas and normalized point with above-mentioned n white wine sample with the sample wine of kind are chosen
Corresponding spectrum data vector y is not obtained1=(y1,1,…,y1,t,…,y1,h) and y2=(y2,1,…,y2,t,…,y2,h), 1≤t
≤ h, the collection of illustrative plates characteristic Similarity value SIM of two sample wine is calculated as follows,According to SIM values
Size determine the similarity height of two sample wine characteristics.
Preferably, the threshold value T=0.1 in the step c.
Preferably, in the step e default remaining spectrum data number h according to the figure of selected different type composition
The number of modal data determines.
Preferably, the default remaining spectrum data number h is determined as the following formula, h=0.8 × A+0.5 × B+0.4 × C+
0.1 × D, wherein A, B, C, D be respectively alcohol component, aldehydes and lactone component in the p shared spectrum datas chosen, acids into
Point and other compositions spectrum data number, p=A+B+C+D.
The present invention by normalize pretreatment can eliminate Wine Sample spectrum data scale differ greatly caused by weight
Difference, pass through low to white wine characteristic contribution rate in principal component analysis removal white wine spectrum data or incoherent spectrum data, sieve
Select the spectrum data big to white wine characteristic contribution rate and participate in Similarity Measure, eliminate that contribution rate is low or incoherent collection of illustrative plates with this
Influence of the data to Similarity value.Contribution rate size of each spectrum data to white wine characteristic is calculated by principal component analysis again,
And then the weights for each white wine spectrum data for participating in Similarity Measure are determined, so as to which the Similarity value for enabling to obtain embodies
Go out contribution rate size of the different characteristic composition to white wine characteristic.The Similarity value calculated using the method for the invention can be more accurate
The similarity degree of white wine characteristic really is embodied, it is more accurate to the objective evaluation of white wine, there is great application value.
Brief description of the drawings
The flow chart of collection of illustrative plates Similarity Measures of Fig. 1 based on white wine characteristic；
The collection of illustrative plates Similarity Measure flow chart based on characteristic of the similar sample wine of Fig. 2 two.
Embodiment
To make the object, technical solutions and advantages of the present invention clearer, below technical scheme in the present invention carry out it is clear
Chu, it is fully described by, it is clear that described embodiment is part of the embodiment of the present invention, rather than whole embodiments.It is based on
Embodiment in the present invention, those of ordinary skill in the art are obtained every other under the premise of creative work is not made
Embodiment, belong to the scope of protection of the invention.
Embodiment 1：
Step carries out the calculating of the collection of illustrative plates Similarity value of certain aromatic Chinese spirit as shown in Figure 1：
(1) selection of white wine sample and spectrum data：
Choose 20 samples of the type white wine, the composition spectrum data of each sample include 43 can obtain it is shared
White wine spectrum data (mg/L), the spectrum data include acetaldehyde (z1), Ethyl formate (z2), ethyl acetate (z3), acetal
(z4), tert-pentyl alcohol (z5), ethyl butyrate (z6), sec-butyl alcohol (z7), normal propyl alcohol (z8), butyl acetate (z9), isobutanol (z10), acetic acid
Isopentyl ester (z11), ethyl valerate (z12), n-butanol (z13), n-amyl acetate (z14), methyl butanol (z15), isoamyl alcohol (z16)、
Ethyl hexanoate (z17), n-amyl alcohol (z18), cognac oil (z19), ethyl lactate (z20), n-hexyl alcohol (z21), ethyl caprilate (z22)、
Furfural (z23), acetic acid (z24), ethyl pelargonate (z25), propionic acid (z26), butyric acid (z27), valeric acid (z28), ethyl butyric acid (z29), caproic acid
(z30), bata-phenethyl alcohol (z31), enanthic acid (z32), octanoic acid (z33), ethyl palmitate (z34), acetone (z35), butanediol (z36), valeric acid
(z37), ethyl linoleate (z38), ethyl palmitate (z39), n-heptanol (z40), 3- hydroxy-2-butanones (z41), ethyl isovalerate
(z42), 3- methylbutyraldehyds (z43)。
(2) white wine spectrum data is normalized：
White wine component content data are normalized, eliminate weight difference, it is 0 to make its average, variance 1.Often
The spectrum data of individual white wine sample is a row vector, and the spectrum data matrix Z of 20 sample compositions is：
Obtain the average u of matrix Z each columnsjWith variances sigmaj, j spans 1≤j≤43.Using equation below to matrix Z's
Each element is standardized：
It is X to obtain normalized matrix：
Relevant data before and after first sample standardization are provided in table 1.
The data of first white wine samples normalization before and after the processing selected by table 1
(3) screens the spectrum data big to white wine characteristic contribution rate by principal component analysis：
Principal component analysis is carried out to data matrix X, obtains matrix X covariance square first
Battle array S：
In formula For the average value of i-th of spectrum data of n Wine Sample.Ask
Go out matrix S eigenvalue λiAnd corresponding orthogonalization unit character vector ai=(ai,1,ai,2,ai,3…ai,p).Given threshold T=
0.1, by the descending arrangement of eigenvalue λ, find eigenvalue λm, make λm>=T, λm+1＜ T, here λ12=0.13, λ13=0.08,
λ12Corresponding principal component is F12, it is by the expression formula of white wine spectrum data：F12=a12,1x1+a12,2x2+...+a12,43x43,
Coefficient a12.1,a12,2,…,a12,43Value it is as shown in table 2：
The principal component of table 2 is F12Coefficient value
From Table 2, it can be seen that principal component F12Coefficient value in a12,30Absolute value with maximum, it is 0.8551, it is corresponding
White wine composition be caproic acid, it is believed that caproic acid is minimum to the contribution rate of white wine characteristic, delete 20 samples in caproic acid composition,
Obtain the matrix X of new 20 × 42：
Repeat the above steps, often perform an above-mentioned steps and delete a white wine composition spectrum data, until each sample
Remaining h composition spectrum data.Here as shown in Table 1, in 43 selected composition spectrum datas, alcohol component 12,
Aldehydes composition 4, lactone component 16, acrylic component 9, other compositions 2, default remaining spectrum data number h=0.8
× 12+0.5 × 20+0.4 × 9+0.1 × 2, is calculated h=23.4, takes h=23.
Remaining h white wine spectrum data y=(y1,y2,…,yh) represent, 23 spectrum datas that each sample is chosen
Form matrix Y：
Choose 23 white wine compositions be respectively：Acetaldehyde (y1), hexanol (y2), methylbutyraldehyd (y3), n-butanol (y4), second
Acetoacetic ester (y5), ethyl lactate (y6), butyric acid (y7), valeric acid (y8), isoamyl alcohol (y9), acetone (y10), furfural (y11), isovaleric acid
Ethyl ester (y12), ethyl hexanoate (y13), normal propyl alcohol (y14), cognac oil (y15), ethyl valerate (y16), isobutanol (y17), methyl
Butanol (y18), ethyl butyrate (y19), acetic acid (y20), propionic acid (y21), butanediol (y22), ethyl oleate (y23)。
(4) obtains the weights when spectrum data screened participates in Similarity Measure by principal component analysis：
Principal component analysis is carried out to matrix Y, k principal component before taking, makes its contribution rate of accumulative total G (k) >=85%, G (k-1) ＜
85%.Here G (k)=83.27%, G (k)=89.83%, k=4 is taken.The contribution rate of preceding 4 principal components is respectively α1=
46.13%, α2=24.35%, α3=12.79%, α4=6.56%.
Preceding 4 principal components and the relation of 23 white wine compositions are as follows：
Consider the relational matrix A of contribution rate0It can be expressed as：
To matrix A0Often row absolute value sums to obtain vectorial b=(b1,b2,…,b23), the element b of the vectorial btUnder
Formula is tried to achieve：bt=| α1a1,t|+|α2a2,t|+|α3a3,t|+|α4a4,t|, btFor the contribution rate of t-th of white wine composition of selection, then it is right
Vectorial b is normalized to obtain vectorial c, element c in ctTried to achieve by following formula:
ct=bt/sum(b)
Contribution rate coefficient vector c=(c after normalization1,c2,…,c23), vectorial c element ctValue be similarity meter
Weights during calculation, its value are as shown in table 3：
Element c in 3 vectorial c of tabletValue
(5) calculating of the Similarity value of two sample wine of：
It is determined that the collection of illustrative plates Similarity value based on white wine characteristic calculates the h needed important spectrum data by foregoing four step
With weights ct., only need to be to 23 collection of illustrative plates of selection as shown in Figure 2 during the collection of illustrative plates Similarity value SIM of two sample wine of step calculating
Data are normalized, and respectively obtain the spectrum data vector y of two sample wine1=(y1,1,y1,2,…y1,23) with
y2=(y2,1,y2,2,…y2,23), and by corresponding weights ctBring formula intoIn calculate
To Similarity value SIM.
Choose 2 sample wine (sample z with above-mentioned 20 white wine samples with kind1、z2) spectrum data carry out similarity
Calculate, table 4 give two sample wine spectrum data and normalization after spectrum data, by the above-mentioned selection of two sample wine
23 white wine compositions normalization after spectrum data bring above-mentioned Similarity value calculation formula into, obtain SIM=0.286,2 institutes
It is smaller to state Similarity value SIM, shows that two sample wine are more close.
Spectrum data before and after the normalized of 4 two sample wine of table
In order to which beneficial effects of the present invention are better described, now choose and sample z described in table 41With 10 samples of kind
Sample wine and (use s1,s2,…,s10Represent), with table 4 described in sample z110 sample wine of different cultivars (use d1,
d2,…,d10Represent), sample s is calculated respectively1,s2,…,s10And d1,d2,…,d10With sample z1Similarity value, phase
Computational methods of the present invention are respectively adopted like the calculating of angle valueCalculating side of the prior art
MethodCarry out, wherein h represents the number of the spectrum data of screening, and p represents unscreened all figures
The number of modal data, obtained Similarity value are as shown in table 5 below.
5 20 Wine Samples of table respectively with the sample z in table 41Similarity value
As can be seen from Table 5：
1st, for the calculating of same breed white wine collection of illustrative plates similarity, the Similarity value that computational methods of the present invention obtain
(average u=0.2082) is significantly lower than the Similarity value (average u=0.5216) that computational methods of the prior art obtain, and this
Invention calculate similarity fluctuation (variances sigma=0.0003) be less than Similarity Measure of the prior art fluctuation (variances sigma=
0.0015) difference of white wine characteristic can more be embodied by, illustrating the data for the Similarity value that the similarity calculating method of the present invention obtains,
And data are more stable；
2nd, for the calculating of different cultivars white wine collection of illustrative plates similarity, the similarity average that the present invention calculates is 0.9170, compared with
Same breed white wine collection of illustrative plates similarity average 0.2082 changes greatly, and illustrates that the similarity calculating method of the present invention objectively can
The difference of white wine characteristic is enough embodied, and the average for the similarity that computational methods of the prior art obtain is 0.5707, it is more identical
The change of kind white wine collection of illustrative plates similarity average 0.5216 is smaller, and this is due to figure low to white wine characteristic contribution rate or without contribution
Modal data participates in calculating, so as to mask embodiment of the similarity of calculating to white wine characteristic.Therefore, collection of illustrative plates similarity of the invention
Computational methods objectively embody the similarity degree of white wine characteristic.
Finally it should be noted that：The above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although
The present invention is described in detail with reference to the foregoing embodiments, it will be understood by those within the art that：It still may be used
To be modified to the technical scheme described in foregoing embodiments, or equivalent substitution is carried out to which part technical characteristic；
And these modification or replace, do not make appropriate technical solution essence depart from various embodiments of the present invention technical scheme spirit and
Scope.
Claims (8)
1. a kind of collection of illustrative plates similarity calculating method based on white wine characteristic, it is characterised in that comprise the following steps：(1) choose same
N different samples of class white wine, p shared spectrum datas are chosen to n sample respectively；The spectrum data refers to white wine composition
Content data；
(2) the p spectrum data is normalized, obtains the collection of illustrative plates number after the p normalization by the n sample
According to the data matrix of composition
<mrow>
<mi>X</mi>
<mo>=</mo>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<msub>
<mi>x</mi>
<mrow>
<mn>1</mn>
<mo>,</mo>
<mn>1</mn>
</mrow>
</msub>
</mtd>
<mtd>
<msub>
<mi>x</mi>
<mrow>
<mn>1</mn>
<mo>,</mo>
<mn>2</mn>
</mrow>
</msub>
</mtd>
<mtd>
<mn>...</mn>
</mtd>
<mtd>
<msub>
<mi>x</mi>
<mrow>
<mn>1</mn>
<mo>,</mo>
<mi>p</mi>
</mrow>
</msub>
</mtd>
</mtr>
<mtr>
<mtd>
<msub>
<mi>x</mi>
<mrow>
<mn>2</mn>
<mo>,</mo>
<mn>1</mn>
</mrow>
</msub>
</mtd>
<mtd>
<msub>
<mi>x</mi>
<mrow>
<mn>2</mn>
<mo>,</mo>
<mn>2</mn>
</mrow>
</msub>
</mtd>
<mtd>
<mn>...</mn>
</mtd>
<mtd>
<msub>
<mi>x</mi>
<mrow>
<mn>2</mn>
<mo>,</mo>
<mi>p</mi>
</mrow>
</msub>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
<mtd>
<mo>.</mo>
</mtd>
<mtd>
<mo>.</mo>
</mtd>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
<mtd>
<mo>.</mo>
</mtd>
<mtd>
<mo>.</mo>
</mtd>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
<mtd>
<mo>.</mo>
</mtd>
<mtd>
<mo>.</mo>
</mtd>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<msub>
<mi>x</mi>
<mrow>
<mi>n</mi>
<mo>,</mo>
<mn>1</mn>
</mrow>
</msub>
</mtd>
<mtd>
<msub>
<mi>x</mi>
<mrow>
<mi>n</mi>
<mo>,</mo>
<mn>2</mn>
</mrow>
</msub>
</mtd>
<mtd>
<mn>...</mn>
</mtd>
<mtd>
<msub>
<mi>x</mi>
<mrow>
<mi>n</mi>
<mo>,</mo>
<mi>p</mi>
</mrow>
</msub>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>;</mo>
</mrow>
(3) principal component analysis is carried out to the spectrum data matrix X, filters out the h collection of illustrative plates number big to white wine characteristic contribution rate
According to h is integer and 1≤h≤P；
(4) data matrix formed to the h spectrum data by n sample carries out principal component analysis, obtains the h figure
The weights of each spectrum data in modal data；
(5) similarity of two Wine Samples is obtained by h spectrum data of screening and corresponding weights.
2. the collection of illustrative plates similarity calculating method according to claim 1 based on white wine characteristic, it is characterised in that the step
(2) comprise the following steps：
A. the p shared spectrum datas are normalized, obtain the spectrum data vector x of each samplei=(xi,1,
xi,2,xi,3,…,xi,p), wherein 1≤i≤n, element xi,pFor the spectrum data after i-th of sample, p-th of normalization；
The above-mentioned spectrum data vector composition data matrix of b.n sampleWherein,
xn,pRepresent the spectrum data after p-th of normalization of n-th of white wine sample.
3. the collection of illustrative plates similarity calculating method according to claim 1 based on white wine characteristic, it is characterised in that the step
(3) comprise the following steps：
C. principal component analysis is carried out to the data matrix X, obtains matrix X covariance matrixWherein elementxk,i、xk,jRespectively
It is i-th and j spectrum data of k-th of sample,It is the average value of i-th and j spectrum data of n sample respectively,
1≤k≤n, 1≤i≤P, 1≤j≤P, obtain matrix S eigenvalue λ and by its descending arrangement, given threshold T, find spy
Value indicative λm, make λm>=T, λm+1＜ T, λmCorresponding orthogonalization unit character vector am=(am,1,…,am,g,…,am,p), λmIt is corresponding
Principal component FmExpression formula by white wine spectrum data is Fm=am,1x1+…+am,gxg+…+am,pxp, wherein am,1,…,
am,g,…,am,pFor FmCharacteristic vector coefficient；
D. above-mentioned principal component F is foundmThe characteristic vector coefficient a of middle maximum absolute valuem,gCorresponding spectrum data xg, delete n in vain
The spectrum data x of each sample in wine sampleg, then each p-1 spectrum data of white wine sample residual；
E. n sample of remaining p-1 spectrum data is repeated the above steps, every time delete a maximum absolute value feature to
Spectrum data corresponding to coefficient of discharge, untill the default h spectrum data of each sample residual.
4. the collection of illustrative plates similarity calculating method according to claim 1 based on white wine characteristic, it is characterised in that the step
(4) comprise the following steps：
F. the spectrum data vector y of each white wine sample is formed with h spectrum data of screeningi=(yi,1,yi,2,yi,3,…,
yi,h), 1≤i≤n, data matrix Y is formed with the above-mentioned spectrum data vector of n sample；
G. principal component analysis is carried out to above-mentioned matrix Y, k principal component before taking, makes its contribution rate of accumulative total G (k) >=85%, G (k-1)
＜ 85%, calculate the contribution rate α of each principal component in k principal componenti, obtain by k principal component and the figure containing h spectrum data
Genealogical relationship matrixWherein, the contribution rate of accumulative total G (k) byCalculate, the contribution rate αiByCalculate；
H. the vectorial b=(b of the contribution margin composition of the h spectrum data screened1,…,bt,…bh), b in formulatFor t-th of collection of illustrative plates number
According to contribution margin, 1≤t≤h, bt=| α1a1,t|+|α2a2,t|+…+|αkak,t|；
I. vectorial b is normalized to obtain normalized contribution rate coefficient vector c=(c1,…,ct,…ch), 1≤t≤h, formula
Middle element ct=bt/ sum (b), it is that rate coefficient, namely t-th of figure during Similarity Measure are contributed in the normalization of t-th of spectrum data
The weights of modal data.
5. the collection of illustrative plates similarity calculating method according to claim 1 based on white wine characteristic, it is characterised in that the step
(5) comprise the following steps：
J. two are chosen to obtain respectively with the h spectrum data and normalized of the sample wine of kind with above-mentioned n white wine sample
To corresponding spectrum data vector y1=(y1,1,…,y1,t,…,y1,h) and y2=(y2,1,…,y2,t,…,y2,h), 1≤t≤h,
The collection of illustrative plates characteristic Similarity value SIM of two sample wine is calculated as follows,According to the big of SIM values
The small similarity height for determining two sample wine characteristics.
6. the collection of illustrative plates similarity calculating method according to claim 3 based on white wine characteristic, it is characterised in that the step
Threshold value T=0.1 in c.
7. the collection of illustrative plates similarity calculating method according to claim 3 based on white wine characteristic, it is characterised in that the step
Default remaining spectrum data number h determines according to the number of the spectrum data of selected different type composition in e.
8. the collection of illustrative plates similarity calculating method according to claim 7 based on white wine characteristic, it is characterised in that described default
Remaining spectrum data number h determine as the following formula, h=0.8 × A+0.5 × B+0.4 × C+0.1 × D, wherein A, B, C, D difference
For the collection of illustrative plates number of alcohol component, aldehydes and lactone component, acrylic component and other compositions in p shared spectrum datas of selection
According to number, p=A+B+C+D.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711277895.3A CN107862348B (en) | 2017-12-06 | 2017-12-06 | Method for calculating similarity of graphs based on characteristics of white spirit |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711277895.3A CN107862348B (en) | 2017-12-06 | 2017-12-06 | Method for calculating similarity of graphs based on characteristics of white spirit |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107862348A true CN107862348A (en) | 2018-03-30 |
| CN107862348B CN107862348B (en) | 2021-12-14 |
Family
ID=61705254
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711277895.3A Active CN107862348B (en) | 2017-12-06 | 2017-12-06 | Method for calculating similarity of graphs based on characteristics of white spirit |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107862348B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109359678A (en) * | 2018-10-09 | 2019-02-19 | 四川理工学院 | A kind of high-precision classification recognizer of white wine map |
| CN109376805A (en) * | 2018-12-21 | 2019-02-22 | 四川理工学院 | A kind of classification method based on white wine base liquor Fingerprints |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080245132A1 (en) * | 2007-04-09 | 2008-10-09 | Head Michael S | Methods of detecting and eliminating tainted cork wine bottle stoppers |
| CN106645254A (en) * | 2016-12-26 | 2017-05-10 | 厦门出入境检验检疫局检验检疫技术中心 | Method for identifying grape variety and year of wine |
| CN106770159A (en) * | 2016-12-02 | 2017-05-31 | 中国计量大学 | High sensitivity food color detection method |
-
2017
- 2017-12-06 CN CN201711277895.3A patent/CN107862348B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080245132A1 (en) * | 2007-04-09 | 2008-10-09 | Head Michael S | Methods of detecting and eliminating tainted cork wine bottle stoppers |
| CN106770159A (en) * | 2016-12-02 | 2017-05-31 | 中国计量大学 | High sensitivity food color detection method |
| CN106645254A (en) * | 2016-12-26 | 2017-05-10 | 厦门出入境检验检疫局检验检疫技术中心 | Method for identifying grape variety and year of wine |
Non-Patent Citations (3)
| Title |
|---|
| 李鸿禧: "基于相关—主成分分析的港口物流评价研究", 《中国优秀硕士学位论文全文数据库经济与管理科学辑》 * |
| 杜玲玲: "金门高粱酒真伪鉴别技术研究", 《中国优秀硕士学位论文全文数据库工程科技Ⅰ辑》 * |
| 牟秉华: "《中国社会经济统计百科全书》", 31 August 1994 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109359678A (en) * | 2018-10-09 | 2019-02-19 | 四川理工学院 | A kind of high-precision classification recognizer of white wine map |
| CN109359678B (en) * | 2018-10-09 | 2022-08-30 | 四川轻化工大学 | High-precision classification recognition algorithm for liquor atlas |
| CN109376805A (en) * | 2018-12-21 | 2019-02-22 | 四川理工学院 | A kind of classification method based on white wine base liquor Fingerprints |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107862348B (en) | 2021-12-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102023137B (en) | Method for identifying white spirits | |
| Schuhfried et al. | Classification of 7 monofloral honey varieties by PTR-ToF-MS direct headspace analysis and chemometrics | |
| US11281895B2 (en) | Expression recognition method, computer device, and computer-readable storage medium | |
| CN113125590B (en) | Objective evaluation method for aroma quality of Yunnan red congou tea soup based on rapid gas-phase electronic nose technology | |
| CN107862348A (en) | A kind of collection of illustrative plates similarity calculating method based on white wine characteristic | |
| CN103226093A (en) | Calibration curve creation method, calibration curve creation device and target component determination device | |
| CN109376805A (en) | A kind of classification method based on white wine base liquor Fingerprints | |
| CN105095652B (en) | Sample component assay method based on stack limitation learning machine | |
| Huang et al. | Rapid and nondestructive determination of sorghum purity combined with deep forest and near-infrared hyperspectral imaging | |
| Porep et al. | Implementation of an on‐line near infrared/visible (NIR/VIS) spectrometer for rapid quality assessment of grapes upon receival at wineries | |
| CN108399433A (en) | A kind of sorting technique based on Dactylogram Chart About Chinese Spirit feature | |
| Sáiz-Abajo et al. | Near infrared spectroscopy and pattern recognition methods applied to the classification of vinegar according to raw material and elaboration process | |
| CN110609011A (en) | Near-infrared hyperspectral detection method and system for starch content of single-kernel corn seeds | |
| CN117332358B (en) | Corn soaking water treatment method and system | |
| CN112036482A (en) | Traditional Chinese medicine classification method based on electronic nose sensor data | |
| CN117874609A (en) | Universal rapid method for rapidly identifying whether natural product is specific production place | |
| Letchov et al. | Growth kinetics of grape berry density (Vitis vinifera L.‘Black Corinth’) | |
| Ma et al. | An intelligent and vision-based system for Baijiu brewing-sorghum discrimination | |
| CN104568639A (en) | Method and device for determination of fruit sugar degree | |
| WO2023229821A1 (en) | Detecting wine characteristics from wine samples | |
| Pusadan et al. | k-Nearest Neighbor and Feature Extraction on Detection of Pest and Diseases of Cocoa | |
| CN116539553A (en) | Method for improving robustness of near infrared spectrum model | |
| CN113935963B (en) | Image recognition detection method and system for litchi embryo development degree | |
| Xie et al. | Ultraviolet spectroscopy method for classifying vinegars | |
| KR101423941B1 (en) | Method for assessing the quality of makgeolli from rice by statistical analysis with visualization procedure |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20190220 Address after: 646000 Airentang Square, Luzhou City, Sichuan Province, China Liquor Golden Triangle Liquor Industry Park Applicant after: Luzhou Laojiao Limited Company Applicant after: Sichuan University of Science & Engineering Address before: 643000 No. 180, Xue Yuan Street, Huxing Road, Zigong, Sichuan. Applicant before: Sichuan University of Science & Engineering |
|
| TA01 | Transfer of patent application right | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |