CN107179294B - A kind of near-infrared spectral analytical method based on subspace analysis - Google Patents
A kind of near-infrared spectral analytical method based on subspace analysis Download PDFInfo
- Publication number
- CN107179294B CN107179294B CN201710638544.4A CN201710638544A CN107179294B CN 107179294 B CN107179294 B CN 107179294B CN 201710638544 A CN201710638544 A CN 201710638544A CN 107179294 B CN107179294 B CN 107179294B
- Authority
- CN
- China
- Prior art keywords
- matrix
- row
- sample
- spectrum
- formula
- 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.)
- Active
Links
- 238000004458 analytical method Methods 0.000 title claims abstract description 23
- 230000003595 spectral effect Effects 0.000 title claims abstract description 11
- 238000001228 spectrum Methods 0.000 claims abstract description 46
- 239000011159 matrix material Substances 0.000 claims description 119
- 238000000034 method Methods 0.000 claims description 19
- 238000002329 infrared spectrum Methods 0.000 claims description 13
- 238000002835 absorbance Methods 0.000 claims description 4
- 238000000354 decomposition reaction Methods 0.000 claims description 4
- 230000017105 transposition Effects 0.000 claims description 4
- 238000000691 measurement method Methods 0.000 claims description 2
- 238000010183 spectrum analysis Methods 0.000 abstract description 5
- 230000007812 deficiency Effects 0.000 abstract description 4
- 238000001237 Raman spectrum Methods 0.000 abstract description 2
- 238000004587 chromatography analysis Methods 0.000 abstract description 2
- 238000005259 measurement Methods 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- LFQSCWFLJHTTHZ-UHFFFAOYSA-N Ethanol Chemical compound CCO LFQSCWFLJHTTHZ-UHFFFAOYSA-N 0.000 description 3
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 229920002472 Starch Polymers 0.000 description 1
- 240000008042 Zea mays Species 0.000 description 1
- 235000005824 Zea mays ssp. parviglumis Nutrition 0.000 description 1
- 235000002017 Zea mays subsp mays Nutrition 0.000 description 1
- 150000001413 amino acids Chemical class 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 150000001875 compounds Chemical class 0.000 description 1
- 239000000470 constituent Substances 0.000 description 1
- 235000005822 corn Nutrition 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 239000003925 fat Substances 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000012628 principal component regression Methods 0.000 description 1
- 102000004169 proteins and genes Human genes 0.000 description 1
- 108090000623 proteins and genes Proteins 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 235000019698 starch Nutrition 0.000 description 1
- 239000008107 starch Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
- G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
- G01N21/359—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light using near infrared light
Landscapes
- Physics & Mathematics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The present invention relates to analytical chemistry field, specially a kind of near-infrared spectral analytical method based on subspace analysis.A kind of near-infrared spectral analytical method based on subspace analysis includes the following steps: (1) calibration set sample treatment；(2) forecast sample group score value is completed to determine.The present invention analyzes sample spectra using unified step, obtains sample component information, overcomes the main deficiency of PLS algorithm.The experimental results showed that estimated performance is better than existing PLS algorithm, predicted root mean square error is respectively less than the predicted root mean square error of PLS.It is also possible to apply the invention to other spectrum analysis, such as ultraviolet spectral analysis, Raman spectrum analysis, chromatography.
Description
Technical field
The present invention relates to analytical chemistry field, specially a kind of near-infrared spectral analytical method based on subspace analysis.
Background technique
Near-infrared spectral analysis technology, which has many advantages, such as that test speed is fast, analytic process is simple, is suitable for diffusing reflection, to be tested,
The technology has been applied to the quick measurement of the constituent contents such as protein, amino acid, fat, starch, moisture.Due near infrared spectrum
With serious plyability, information relevant to component content is difficult directly to extract and give to close in substance near infrared spectrum
The spectrum resolution of reason.Therefore, it is necessary to effective letter is extracted from the spectrum of measurement using the multivariate calibration methods in Chemical Measurement
Breath carries out quantitative analysis.
Traditional multivariate calibration methods mainly have K matrix method, P matrix method, multi-element linear regression method, principal component regression
Analytic approach, Partial Least Squares (PLS) etc..Many experiments show that PLS algorithm performance is better than other sides in most cases
Method.Although PLS algorithm has a large amount of application, PLS algorithm is primarily present following deficiency: (1) PLS algorithm calculates score matrix
With need to consider simultaneously when loading matrix spectrum and group sub-matrix, if predicted in group sub-matrix there are unknown component information
As a result it will be affected；(2) determination of hidden variable is there are many method in PLS algorithm, the hidden variable for selecting distinct methods to determine,
Its prediction result difference is larger.
Summary of the invention
The present invention is directed to overcome the main deficiency of PLS algorithm, a kind of near infrared spectrum based on subspace analysis point is proposed
Analysis method.
Technical program of the present invention lies in:
A kind of near-infrared spectral analytical method based on subspace analysis, includes the following steps:
(1) calibration set sample treatment:
Step 1.1 selects p representative samples as calibration set sample, and wherein p is positive integer；It is close according to sample
Infrared spectrum response property measures its near infrared spectrum in M wavelength, is measured to sample, obtains its N number of group of score value；
The near infrared light spectrum matrix X that step 1.2 p row, M are arranged indicates the near infrared spectrum of calibration set sample, and wherein X's is every
A line represents the absorbance of a sample；By the group sub-matrix Y that same sequential build p row, N are arranged, wherein every a line of Y represents one
The group score value of a sample；
Step 1.3 near infrared light spectrum matrix X pretreatment:
Step 1.3.1 calculates the average value row vector mX of the near infrared light spectrum matrix X constituted in above-mentioned steps 1.2, as follows
It is described:
In formula, mXkFor the kth column element of row vector mX, Xi,kFor the i-th row k column element of matrix X；
Mean value near infrared light spectrum matrix X0 is removed in step 1.3.2 calculating
X0I, k=XI, k-mXkI=1,2 ..., p (2)
In formula, X0i,kFor the i-th row k column element of matrix X0, Xi,kFor the i-th row k column element of matrix X；
Step 1.4 group sub-matrix Y pretreatment:
Step 1.4.1 calculates the average value row vector mY of the group sub-matrix Y constituted in above-mentioned steps 1.2, as described below:
In formula, mYkFor the kth column element of row vector mY, Yi,kFor the i-th row k column element of matrix Y
Mean value group sub-matrix Y0 is removed in step 1.4.2 calculating, as described below:
Y0I, k=YI, k-mYkI=1,2 ..., p (4)
In formula, Y0i,kFor the i-th row k column element of matrix Y0；
Step 1.5 determines subspace:
Step 1.5.1 calculates the autocorrelation matrix R0 for removing mean value near infrared light spectrum matrix X0, as described below:
In formula, subscript T representing matrix transposition；
The eigenvalue λ of step 1.5.2 calculating autocorrelation matrix R0i(i=1,2 ..., p) and sequence make λp≥…≥λ2≥λ1
=0；
Step 1.5.3 determines noise subspace dimension n n:
Level of significance α (α≤0.05) is arranged in step 1.5.3.1, calculates standardized normal distribution upper percentage point Nα；
The initial dimension n n (nn >=p/4) of noise subspace is arranged in step 1.5.3.2；
The average value ms of step 1.5.3.3 calculating section characteristic value, as described below:
Step 1.5.3.4 calculates eigenvalue λiMean square deviation
If step 1.5.3.5 λnn+1≤ms+ss×Nα, nn=nn+1 is enabled, 1.5.3.3 is gone to step, otherwise goes to step 1.6；
Step 1.6 output parameter X0, mX, Y0, mY, nn；
(2) forecast sample group score value is completed to determine:
Step 2.1 measures forecast sample near infrared spectrum by calibration set sample near-infrared spectral measurement method, and is denoted as M column
Vector x t；
Step 2.2 pre-processes forecast sample spectrum, obtains spectrum x0, as described below:
X0=xt-mX (8)；
Step 2.3 constructs extended matrix Xs:
The autocorrelation matrix R of step 2.4 calculating extended matrix Xs
Step 2.5 carries out Eigenvalues Decomposition to matrix R
R=VTDV (11)
In formula, matrix D is with vector [d1,d2,…,dp+1] be diagonal element diagonal matrix, and d1≤d2≤…≤
dp+1；
The matrix V n that the preceding nn column of step 2.6 order matrix V are constituted is noise subspace；
Vn piecemeal is two matrix Vs 1 and V0 by step 2.7, wherein V1 is the matrix that the preceding p row of Vn is constituted, and V0 is Vn
The matrix that+1 row of pth is constituted；
Step 2.8 calculates initial estimate, as described below:
Step 2.9 output estimation result y, as described below:
Y=y0+Y0 (13).
The technical effects of the invention are that:
The present invention analyzes sample spectra using unified step, obtains sample component information, overcomes PLS algorithm
Main insufficient, the experimental results showed that estimated performance is better than existing PLS algorithm, predicted root mean square error is respectively less than the prediction of PLS
Root-mean-square error.It is also possible to apply the invention to other spectrum analysis, such as ultraviolet spectral analysis, Raman spectrum analysis, chromatography.
Detailed description of the invention
Fig. 1 is that the present invention is based on the flow charts of the near-infrared spectral analytical method of subspace analysis.
Specific embodiment
Embodiment 1:
Prepare alcohol water blend, alcohol concentration range be 0.02~0.61 (volume ratio, interval 0.01), totally 60 samples,
Adjacent concentration difference 0.01.These samples are successively put into 70 Fourier Transform Near Infrared instrument of Brooker company VERTEX
Upper its near infrared spectrum of measurement, wave-length coverage are 770nm~1300nm (totally 2593 wavelength, i.e. M=2593).It is by concentration
0.03,0.06,0.09 ..., 0.6 sample (totally 20) is used as forecast sample, remaining sample (totally 40, i.e. p=40) is constituted
Calibration set sample spectrum.Calibration set sample spectrum is denoted as matrix X, concentration is denoted as the matrix Y of 40 rows 1 column.Prediction specific implementation
Step is as follows with feature:
Step 1 calibration set sample treatment:
Step 1.1 compound concentration is 0.02~0.61 (volume ratio, the alcohol water blend of interval 0.01), in Brooker company
The near infrared spectrum that wave-length coverage is 770nm~1300nm is measured in 70 Fourier Transform Near Infrared instrument of VERTEX, is chosen
Concentration is 0.03,0.06,0.09 ..., and for the sample at 0.6 (interval 0.03) as forecast sample, remaining sample is calibration set sample.
Step 1.2 p (=40) row, M (=2593) column matrix X indicate calibration set sample measure spectrum, and wherein X's is each
Row represents the absorbance of a sample；It goes by same sequential build p (=40), N (=1) column matrix Y, wherein every a line generation of Y
The concentration of one sample of table.
Step 1.3 Pretreated spectra:
Step 1.3.1 is calculated the average value row vector mX of spectrum matrix X by formula (1)
In formula, mXkFor the kth column element of row vector mX, Xi,kFor the i-th row k column element of matrix X
Step 1.3.2 is calculated by formula (2) and is removed mean value spectrum matrix X0
X0I, k=XI, k-mXkI=1,2 ..., p (2)
In formula, X0i,kFor the i-th row k column element of matrix X0, Xi,kFor the i-th row k column element of matrix X
Step 1.4 group sub-matrix Y pretreatment:
Step 1.4.1 presses the average value row vector mY of formula (3) calculating group sub-matrix
In formula, mYi,kFor the kth column element of row vector mY, Yi,kFor the i-th row k column element of matrix Y
Step 1.4.2 is calculated by formula (4) and is removed mean value group sub-matrix Y0
Y0I, k=YI, k-mYkI=1,2 ..., p (4)
In formula, Y0i,kFor the i-th row k column element of matrix Y0.
Step 1.5 determines subspace:
Step 1.5.1 is calculated the autocorrelation matrix R0 of spectrum matrix X0 by formula (5)
In formula, subscript T representing matrix transposition；
The eigenvalue λ of step 1.5.2 calculating autocorrelation matrix R0i(i=1,2 ..., p) and sequence make λp≥…≥λ2≥λ1
=0
Step 1.5.3 determines noise subspace dimension n n:
Level of significance α=0.015 is arranged in step 1.5.3.1, calculates standardized normal distribution upper percentage point Nα
The initial dimension n n=13 of noise subspace is arranged in step 1.5.3.2
Step 1.5.3.3 calculates characteristic value average value by formula (6)
Step 1.5.3.4 calculates characteristic value mean square deviation by formula (7)
If step 1.5.3.5 λnn+1≤ms+ss×Nα, nn=nn+1 is enabled, 1.5.3.3 is gone to step, otherwise goes to step 1.6；
Step 1.6 output parameter X0, mX, Y0, mY, nn.
Step 2 forecast sample group score value determines:
Step 2.1 is denoted as M (=2593) dimension row vector xt to each forecast set sample spectra.
Step 2.2 pre-processes forecast sample spectrum by formula (8), obtains spectrum x0
X0=xt-mX (8)
Step 2.3 constructs extended matrix Xs by formula (9)
Step 2.4 is calculated the autocorrelation matrix R of extended matrix Xs by formula (10)
Step 2.5 carries out Eigenvalues Decomposition to matrix R
R=VTDV (11)
In formula, matrix D is with vector [d1,d2,…,dp+1] be diagonal element diagonal matrix, and d1≤d2≤…≤
dp+1；
The matrix V n that the preceding nn column of step 2.6 order matrix V are constituted is noise subspace
Vn piecemeal is two matrix Vs 1 and V0 by step 2.7, wherein V1 is the matrix that the preceding p row of Vn is constituted, and V0 is Vn
The row vector that+1 row of pth is constituted
Step 2.8 calculates initial estimate by formula (12)
Step 2.9 presses formula (13) output estimation result y
Y=y0+Y0 (13).
For convenience of comparison, same data are modeled and predicted using PLS algorithm, and uses predicted root mean square error
RMSEP assesses estimated performance.RMSEP is defined as herein
In formula, ykFor the prediction result of k-th of sample, ytkFor the legitimate reading of k-th of sample, Q is forecast sample sum
20。
Spectrum and the normalized pretreatment of component in the prediction of PLS algorithm, using staying a crosscheck method to carry out hidden change
Measure number selection.It is as shown in table 1 with PLS prediction result using the present invention:
The comparison of 1 embodiment of table, 1 prediction result
| Prediction technique | The present invention | PLS |
| RMSEP | 0.004695 | 0.005756 |
Embodiment 2:
It analyzes near infrared spectrum data is disclosed on internet, data network address are as follows: http: //
www.eigenvector.com/data/Corn/index.html.Select m5spec spectrum as test spectrum when analysis.This
A spectrum shares 80 samples, and wave-length coverage is 1100~2498nm (totally 700 wavelength, i.e. M=700).Sample contains 4 components
Information (i.e. N=4).In the present embodiment, sample is divided by correction and test set using DUPLEX method, wherein test set number is
[2,4,7,8,16,29,31,33,37,38,42,44,45,46,50,52,61,63,70,74], remaining is calibration set (i.e. p=
60).It is finally obtained the result is that calibration set spectrum matrix X is 60 row, 700 column matrix, group sub-matrix be Y is 60 row, 4 column matrix.
Specific implementation step is as follows with feature:
Step 1 calibration set sample treatment:
Step 1.1 be by number [2,4,7,8,16,29,31,33,37,38,42,44,45,46,50,52,61,63,70,
74] for sample as forecast set, remaining sample reads calibration set sample spectrum and group score value as forecast set.
Step 1.2 indicates calibration set sample measure spectrum with p (=60) row M (=700) column matrix X, wherein every a line of X
Represent the absorbance of a sample；By same sequential build p (=60) row N (=4) column matrix Y, wherein every a line of Y represents one
4 group score values of a sample.
Step 1.3 Pretreated spectra:
Step 1.3.1 is calculated the average value row vector mX of spectrum matrix X by formula (1)
In formula, mXkFor the kth column element of row vector mX, Xi,kFor the i-th row k column element of matrix X
Step 1.3.2 is calculated by formula (2) and is removed mean value spectrum matrix X0
X0I, k=XI, k-mXkI=1,2 ..., p (2)
In formula, X0i,kFor the i-th row k column element of matrix X0, Xi,kFor the i-th row k column element of matrix X
The pretreatment of step 1.4 component:
Step 1.4.1 presses the average value row vector mY of formula (3) calculating group sub-matrix
In formula, mYkFor the kth column element of row vector mY, Yi,kFor the i-th row k column element of matrix Y
Step 1.4.2 is calculated by formula (4) and is removed mean value group sub-matrix Y0
Y0I, k=YI, k-mYkI=1,2 ..., p (4)
In formula, Y0i,kFor the i-th row k column element of matrix Y0.
Step 1.5 determines subspace:
Step 1.5.1 is calculated the autocorrelation matrix R0 of spectrum matrix X0 by formula (5)
In formula, subscript T representing matrix transposition
The eigenvalue λ of step 1.5.2 calculating autocorrelation matrix R0i(i=1,2 ..., p) and sequence make λp≥…≥λ2≥λ1
=0
Step 1.5.3 determines noise subspace dimension n n:
Level of significance α=0.015 is arranged in step 1.5.3.1, calculates standardized normal distribution upper percentage point Nα
The initial dimension n n=20 of noise subspace is arranged in step 1.5.3.2
Step 1.5.3.3 calculates characteristic value average value by formula (6)
Step 1.5.3.4 calculates characteristic value mean square deviation by formula (7)
If step 1.5.3.5 λnn+1≤ms+ss×Nα, nn=nn+1 is enabled, 1.5.3.3 is gone to step, otherwise goes to step 1.6；
Step 1.6 output parameter X0, mX, Y0, mY, nn.
Step 2 forecast sample group score value determines:
Step 2.1 is denoted as M (=700) column vector xt to each forecast set sample spectra.
Step 2.2 pre-processes forecast sample spectrum by formula (8), obtains spectrum x0
X0=xt-mX (8)
Step 2.3 constructs extended matrix Xs by formula (9)
Step 2.4 is calculated the autocorrelation matrix R of extended matrix Xs by formula (10)
Step 2.5 carries out Eigenvalues Decomposition to matrix R
R=VTDV (11)
In formula, matrix D is with vector [d1,d2,…,dp+1] be diagonal element diagonal matrix, and d1≤d2≤…≤
dp+1；
The matrix V n that the preceding nn column of step 2.6 order matrix V are constituted is noise subspace
Vn piecemeal is two matrix Vs 1 and V0 by step 2.7, wherein V1 is the matrix that the preceding p row of Vn is constituted, and V0 is Vn
The row vector that+1 row of pth is constituted
Step 2.8 calculates initial estimate by formula (12)
Step 2.9 presses formula (13) output estimation result y
Y=y0+Y0 (13).
For convenience of comparison, same data are modeled and predicted using PLS algorithm, and uses predicted root mean square error
RMSEP assesses estimated performance.RMSEP is defined as herein
In formula, RMSEPiFor the predicted root mean square error of i-th of component, yK, iFor i-th of component prediction of k-th of sample
As a result, ytK, iFor i-th of component legitimate reading of k-th of sample, Q is forecast sample sum 20.
Spectrum and the normalized pretreatment of component in the prediction of PLS algorithm, using staying a crosscheck method to carry out hidden change
Measure number selection.Table 2 is the comparison of this method and PLS predicted root mean square error.
2 embodiment of table, 2 prediction result contrast table
| Component | The present invention | PLS |
| 1 | 0.0077 | 0.0482 |
| 2 | 0.0309 | 0.0829 |
| 3 | 0.0712 | 0.1138 |
| 4 | 0.0822 | 0.3092 |
It can be seen that the prediction root mean square mistake that predicted root mean square error of the invention is respectively less than PLS from table 1,2 result of table
Difference, the present invention are a kind of near-infrared spectral analytical methods of function admirable.Method divides sample spectra using unified step
Analysis, and sample component information is obtained, overcome the main deficiency of PLS algorithm.
Claims (1)
1. a kind of near-infrared spectral analytical method based on subspace analysis, characterized by the following steps:
(1) calibration set sample treatment:
Step 1.1 selects p representative samples as calibration set sample, and wherein p is positive integer；It is closely red according to sample
External spectrum response property measures its near infrared spectrum in M wavelength, is measured to sample, obtains its N number of group of score value；
The near infrared light spectrum matrix X that step 1.2 p row, M are arranged indicates the near infrared spectrum of calibration set sample, and wherein X's is each
Row represents the absorbance of a sample；By the group sub-matrix Y that same sequential build p row, N are arranged, wherein every a line of Y represents one
The group score value of sample；
Step 1.3 near infrared light spectrum matrix X pretreatment:
Step 1.3.1 calculates the average value row vector mX of the near infrared light spectrum matrix X constituted in above-mentioned steps 1.2, following institute
It states:
In formula, mXkFor the kth column element of row vector mX, Xi,kFor the i-th row k column element of matrix X；
Mean value near infrared light spectrum matrix X0 is removed in step 1.3.2 calculating
X0I, k=XI, k-mXkI=1,2 ..., p (2)；
In formula, X0i,kFor the i-th row k column element of matrix X0, Xi,kFor the i-th row k column element of matrix X；
Step 1.4 group sub-matrix Y pretreatment:
Step 1.4.1 calculates the average value row vector mY of the group sub-matrix Y constituted in above-mentioned steps 1.2, as described below:
In formula, mYkFor the kth column element of row vector mY, Yi,kFor the i-th row k column element of matrix Y
Mean value group sub-matrix Y0 is removed in step 1.4.2 calculating, as described below:
Y0I, k=YI, k-mYkI=1,2 ..., p (4)；
In formula, Y0i,kFor the i-th row k column element of matrix Y0；
Step 1.5 determines subspace:
Step 1.5.1 calculates the autocorrelation matrix R0 for removing mean value near infrared light spectrum matrix X0, as described below:
In formula, subscript T representing matrix transposition；
The eigenvalue λ of step 1.5.2 calculating autocorrelation matrix R0i, wherein i=1,2 ..., p, and sort and make λp≥…≥λ2≥λ1
=0；
Step 1.5.3 determines noise subspace dimension n n:
Level of significance α is arranged in step 1.5.3.1, and α≤0.05 calculates standardized normal distribution upper percentage point Nα；
The initial dimension n n of noise subspace is arranged in step 1.5.3.2, wherein nn >=p/4；
The average value ms of step 1.5.3.3 calculating section characteristic value, as described below:
Step 1.5.3.4 calculates eigenvalue λiMean square deviation
If step 1.5.3.5 λnn+1≤ms+ss×Nα, nn=nn+1 is enabled, 1.5.3.3 is gone to step, otherwise goes to step 1.6；
Step 1.6 output parameter X0, mX, Y0, mY, nn；
(2) forecast sample group score value is completed to determine:
Step 2.1 by calibration set sample near-infrared spectral measurement method measure forecast sample near infrared spectrum, and be denoted as M arrange to
Measure xt；
Step 2.2 pre-processes forecast sample spectrum, obtains spectrum x0, as described below:
X0=xt-mX (8)；
Step 2.3 constructs extended matrix Xs:
The autocorrelation matrix R of step 2.4 calculating extended matrix Xs
Step 2.5 carries out Eigenvalues Decomposition to matrix R
R=VTDV (11)；
In formula, matrix D is with vector [d1,d2,…,dp+1] be diagonal element diagonal matrix, and d1≤d2≤…≤dp+1；
The matrix V n that the preceding nn column of step 2.6 order matrix V are constituted is noise subspace；
Vn piecemeal is two matrix Vs 1 and V0 by step 2.7, wherein V1 is the matrix that the preceding p row of Vn is constituted, and V0 is the pth of Vn
The matrix that+1 row is constituted；
Step 2.8 calculates initial estimate, as described below:
Step 2.9 output estimation result y, as described below:
Y=y0+Y0 (13).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710638544.4A CN107179294B (en) | 2017-07-31 | 2017-07-31 | A kind of near-infrared spectral analytical method based on subspace analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710638544.4A CN107179294B (en) | 2017-07-31 | 2017-07-31 | A kind of near-infrared spectral analytical method based on subspace analysis |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107179294A CN107179294A (en) | 2017-09-19 |
| CN107179294B true CN107179294B (en) | 2019-09-03 |
Family
ID=59838184
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710638544.4A Active CN107179294B (en) | 2017-07-31 | 2017-07-31 | A kind of near-infrared spectral analytical method based on subspace analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107179294B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103018178A (en) * | 2012-12-06 | 2013-04-03 | 江苏省质量安全工程研究院 | LPP (Local Preserving Projection)-based Infrared spectrometer calibration method |
| CN105761272A (en) * | 2016-03-16 | 2016-07-13 | 北京航空航天大学 | Pure substance quantity determination method in imaging spectrum mixed pixels |
| CN105784626A (en) * | 2016-04-05 | 2016-07-20 | 中国科学院合肥物质科学研究院 | Atmospheric pollutant self-adaptive identification method and system based on infrared spectrum imaging technology |
-
2017
- 2017-07-31 CN CN201710638544.4A patent/CN107179294B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103018178A (en) * | 2012-12-06 | 2013-04-03 | 江苏省质量安全工程研究院 | LPP (Local Preserving Projection)-based Infrared spectrometer calibration method |
| CN105761272A (en) * | 2016-03-16 | 2016-07-13 | 北京航空航天大学 | Pure substance quantity determination method in imaging spectrum mixed pixels |
| CN105784626A (en) * | 2016-04-05 | 2016-07-20 | 中国科学院合肥物质科学研究院 | Atmospheric pollutant self-adaptive identification method and system based on infrared spectrum imaging technology |
Non-Patent Citations (4)
| Title |
|---|
| A novel variable selection approach that iteratively optimizes variable space using weighted binary matrix sampling;Bai chuan Deng;《Analyst》;20140716;4836–4845 |
| A simple method for multivariate calibration with minimization of the prediction relative error;Xue mei Wu;《Analytical Methods》;20140330;4056–4060 |
| 基于PLS 投影分析的光谱波段选择方法;淡图南等;《光谱学与光谱分析》;20090228;351-354 |
| 基于一元线性回归的近红外光谱模型传递研究;杨辉华等;《分析化学研究报告》;20140930;1229 ～ 1234 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107179294A (en) | 2017-09-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Bázár et al. | NIR detection of honey adulteration reveals differences in water spectral pattern | |
| Hu et al. | Rapid detection of three quality parameters and classification of wine based on Vis-NIR spectroscopy with wavelength selection by ACO and CARS algorithms | |
| Cozzolino et al. | Analysis of grapes and wine by near infrared spectroscopy | |
| Shao et al. | Visible/near-infrared spectra for linear and nonlinear calibrations: a case to predict soluble solids contents and pH value in peach | |
| Chen et al. | Determination of protein, total carbohydrates and crude fat contents of foxtail millet using effective wavelengths in NIR spectroscopy | |
| Büning-Pfaue | Analysis of water in food by near infrared spectroscopy | |
| US20100297291A1 (en) | Visible/near-infrared spectrum analyzing method and grape fermenting method | |
| CN109669023A (en) | A kind of soil attribute prediction technique based on Multi-sensor Fusion | |
| Fadock et al. | Visible-near infrared reflectance spectroscopy for nondestructive analysis of red wine grapes | |
| KR100921914B1 (en) | Component measuring apparatus for grain and measuring method using the same | |
| Yu et al. | Quality determination of Chinese rice wine based on Fourier transform near infrared spectroscopy | |
| Beghi et al. | Rapid monitoring of grape withering using visible near‐infrared spectroscopy | |
| CN102313712B (en) | Correction method of difference between near-infrared spectrums with different light-splitting modes based on fiber material | |
| Fernández-Novales et al. | A feasibility study on the use of a miniature fiber optic NIR spectrometer for the prediction of volumic mass and reducing sugars in white wine fermentations | |
| KR20150112902A (en) | Simultaneous analysis method for content of nutritional components in various foods with different matrix and form using near-infrared reflectance spectroscopy | |
| CN108107019A (en) | A kind of method that versicolorin content in corn is quickly detected based near infrared spectroscopy | |
| Jiang et al. | Molecular spectroscopic wavelength selection using combined interval partial least squares and correlation coefficient optimization | |
| CN107036999A (en) | A kind of five yuan of ready-mixed oil quantitative analysis methods based near infrared spectrum and Chemical Measurement | |
| Zhu et al. | Prediction of wood property in Chinese Fir based on visible/near-infrared spectroscopy and least square-support vector machine | |
| CN107179294B (en) | A kind of near-infrared spectral analytical method based on subspace analysis | |
| Wang et al. | Non-destructive quantitative analysis of carotene content in carrots using Raman spectroscopy | |
| Shen et al. | Evaluation of near‐infrared and mid‐infrared spectroscopy for the determination of routine parameters in Chinese rice wine | |
| CN112763448A (en) | ATR-FTIR technology-based method for rapidly detecting content of polysaccharides in rice bran | |
| CN105866060A (en) | Method for quickly measuring content of thickener in milk beverage through near-infrared spectrometry | |
| Weindorf et al. | Advanced modeling of soil biological properties using visible near infrared diffuse reflectance spectroscopy |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |