CN108548794A - A kind of biological products method for transferring near infrared model - Google Patents
A kind of biological products method for transferring near infrared model Download PDFInfo
- Publication number
- CN108548794A CN108548794A CN201810339983.XA CN201810339983A CN108548794A CN 108548794 A CN108548794 A CN 108548794A CN 201810339983 A CN201810339983 A CN 201810339983A CN 108548794 A CN108548794 A CN 108548794A
- Authority
- CN
- China
- Prior art keywords
- model
- spectrum
- sample
- nir
- near infrared
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 85
- 238000012546 transfer Methods 0.000 claims abstract description 52
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 27
- LFQSCWFLJHTTHZ-UHFFFAOYSA-N Ethanol Chemical compound CCO LFQSCWFLJHTTHZ-UHFFFAOYSA-N 0.000 claims abstract description 25
- 210000002381 plasma Anatomy 0.000 claims abstract description 18
- 238000001556 precipitation Methods 0.000 claims abstract description 16
- 238000004519 manufacturing process Methods 0.000 claims abstract description 12
- 210000004369 blood Anatomy 0.000 claims abstract description 9
- 239000008280 blood Substances 0.000 claims abstract description 9
- 102000004169 proteins and genes Human genes 0.000 claims abstract description 6
- 108090000623 proteins and genes Proteins 0.000 claims abstract description 6
- 238000001228 spectrum Methods 0.000 claims description 76
- 230000003595 spectral effect Effects 0.000 claims description 19
- 238000004458 analytical method Methods 0.000 claims description 12
- 238000002329 infrared spectrum Methods 0.000 claims description 12
- 238000012795 verification Methods 0.000 claims description 12
- 238000012937 correction Methods 0.000 claims description 10
- 238000004497 NIR spectroscopy Methods 0.000 claims description 8
- 230000005540 biological transmission Effects 0.000 claims description 6
- 238000005259 measurement Methods 0.000 claims description 6
- 238000011156 evaluation Methods 0.000 claims description 5
- 239000011521 glass Substances 0.000 claims description 5
- 210000005239 tubule Anatomy 0.000 claims description 5
- 230000000694 effects Effects 0.000 claims description 4
- 230000010354 integration Effects 0.000 claims description 3
- 230000001419 dependent effect Effects 0.000 claims description 2
- 238000007781 pre-processing Methods 0.000 claims description 2
- 238000005516 engineering process Methods 0.000 abstract description 9
- 238000003556 assay Methods 0.000 abstract description 4
- 238000011160 research Methods 0.000 abstract description 4
- 238000010238 partial least squares regression Methods 0.000 abstract description 3
- 241001347978 Major minor Species 0.000 abstract description 2
- 238000003908 quality control method Methods 0.000 abstract description 2
- 235000019441 ethanol Nutrition 0.000 description 18
- 230000008569 process Effects 0.000 description 8
- 239000000047 product Substances 0.000 description 7
- 239000000243 solution Substances 0.000 description 6
- 238000004611 spectroscopical analysis Methods 0.000 description 6
- 239000011159 matrix material Substances 0.000 description 5
- 238000002835 absorbance Methods 0.000 description 4
- 238000002474 experimental method Methods 0.000 description 4
- 238000006243 chemical reaction Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 3
- 239000000463 material Substances 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000009499 grossing Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000011057 process analytical technology Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 239000006228 supernatant Substances 0.000 description 2
- 230000007704 transition Effects 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 238000000862 absorption spectrum Methods 0.000 description 1
- 239000008351 acetate buffer Substances 0.000 description 1
- 239000008186 active pharmaceutical agent Substances 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- OHJMTUPIZMNBFR-UHFFFAOYSA-N biuret Chemical compound NC(=O)NC(N)=O OHJMTUPIZMNBFR-UHFFFAOYSA-N 0.000 description 1
- 239000010836 blood and blood product Substances 0.000 description 1
- 229940125691 blood product Drugs 0.000 description 1
- 238000001831 conversion spectrum Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 229940079593 drug Drugs 0.000 description 1
- 230000005670 electromagnetic radiation Effects 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000004615 ingredient Substances 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000001320 near-infrared absorption spectroscopy Methods 0.000 description 1
- 238000002203 pretreatment Methods 0.000 description 1
- -1 protein Chemical compound 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000010187 selection method Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
- 238000010183 spectrum analysis Methods 0.000 description 1
- 230000006641 stabilisation Effects 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 239000002699 waste material 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 biological blood article manufacturing process Quality Control Technology field is belonged to, it is related to a kind of biological products method for transferring near infrared model.The present invention uses Rank SPXY calibration samples collection division methods, on this basis, the PRS algorithms of binding model transfer realize different resolution different wave length and count the Model transfer of major-minor instrument, establish Partial Least Squares Regression (partial least squares regression, PLSR) model for quantitative forecast.The Model transfer strategy of Rank SPXY PRS is formd, and is used successfully to the assay research of total protein during biological products production blood plasma alcohol precipitation.
Description
Technical field
The present invention relates to biological blood article manufacturing process Quality Control Technology field is belonged to, it is close to be related to a kind of biological products
Infrared Model transfer method.
Background technology
Near infrared spectrum (NearInfraredSpectroscopy, NIRS) be between visible light (Vis) and in it is infrared
(MIR) near infrared spectrum is defined as 780-2526nm's by the electromagnetic radiation as waves between, U.S. material detection association (ASTM)
Region is first non-visible light area that people have found in absorption spectrum.As widely used process analysis technique
(ProcessAnalyticalTechnologies, PAT) tool, near-infrared spectral analysis technology is due to quick nondestructive, environment friend
The features such as good type, in pharmacy, food, agriculture field has potential application, it can be achieved that assay, qualitative point lossless
Analysis, is detected, the functions such as On-line Control more at the scene.The flow that is normally applied of near infrared spectroscopic method is spectra collection, spectrum
The foundation of pretreatment and model.Ft-nir spectrometer (FT-NIR) is the close red of common analytic type in laboratory
The advantages that external spectrum instrument, there is that accuracy is high, reproducible, high resolution, however corresponding to be higher equipment make
With and maintenance cost, and it is more difficult be applied to Site Detection.In recent years, as Site Detection, online production process conditions are wanted
It asks, the near infrared spectroscopy instrument of minitype portable is quickly grown, such as miniature near infrared spectrometer (MicroNIR1700), master
It is exactly small to want feature, and price is relatively inexpensive, and sensitivity is low compared with FT-NIR, and wavelength band is narrow, and accuracy is relatively almost.
The process of establishing of one steady model is sufficiently complex and cumbersome, and with high costs.
The Model transfer technology (calibrationtransfertechnology) of near-infrared is in modern near infrared spectrum
For a new technology growing up of applicability for model is continuously improved in the application process of analytical technology, the purpose is to will be
Steady be transplanted on other same types or different types of instrument of calibration model established on one instrument uses, to improve
The applicability of model or the precision for improving instrument institute established model on the same stage.It is well known that during spectra collection, the stabilization of instrument
For property there are multiple influence factors, these influence factors will directly influence the applicable performance of model.Near-infrared correction is established simultaneously
The chemical score or fundamental property for generally requiring to measure great amount of samples when model are used as data basis, input greatly, of high cost, therefore make
It is shared and is efficiently used very necessary with Model transfer technology implementation model.Model transfer can overcome sample on different instruments
Inconsistency between measurement signal (or spectrum) not only makes by signal processing to eliminate influence of the instrument to measurement signal
There is model that there is preferable dynamic adaptable, and human and material resources, financial resources and time caused by repeating to model can be reduced
Waste.And previously more researchs are all the instrument models transfers of same producer, because instrument interior structure is similar, difficulty is not
Greatly.The Model transfer research of different manufacturers instrument is related to less, and internal structure gap is big, and the sample size after transfer for verification is very big
Etc. difficult points.This is also the reason of user does not change producer generally easily, these nearest problems have caused the extensive concern of people.
Model transfer is that analytic type near infrared spectrometer with high precision is main instrument (masterinstrument), so as to
Take instrument (salveinstrument) supplemented by the type near infrared spectrometer that declines.Model transfer algorithm mainly has 2 major class at present:Have
Standard specimen algorithm and Unmarked word algorithm.Have standard specimen algorithm that a certain number of samples must be selected to constitute standard specimen collection, at the same in main instrument and
The spectral signal for measuring sample respectively from instrument, finds out functional relation, as DS (directstandardization) algorithm,
PDS (piecewisedirectstandardization) algorithm, PRS (piecewise reverse
Standardization) algorithm and Shenk's algorithms etc..Unmarked word algorithm does not need any sample, mainly with FIR
(finiteimpulseresponse) algorithm is representative.It can be reliable by the model established on an instrument by these algorithms
Be transplanted in other instruments and use, or will be a certain under the conditions of the model established be suitable under the other conditions of instrument on the same stage
Spectroscopic data, while spectroscopic data is constantly corrected by this algorithm, the exceptional value in sample is removed, instrument variation and ring are eliminated
Then the drift of instrument spectral data caused by the factors such as border again models revised spectroscopic data, to improve model
Prediction accuracy and precision.Present document relates to Model transfer algorithm be PRS, be based on PDS algorithms rewrite, i.e., by main instrument
(FT-NIR) spectrum translates into the spectrum of auxiliary instrument (Micro NIR 1700).Due to the number of wavelengths and number between two kinds of instruments
Strong point number is different, before Model transfer, needs to choose the transfer of calibration samples participation spectrum and modeling appropriate.Ideal calibration set
Following condition should be met：(1) sample spectrum feature and property ranges should be able to cover the spectral signature of unknown sample；(2) sample
This physico-chemical parameter should be equally distributed.But in practical measurement process, the component content distribution of sample is not uniformly to divide
Cloth, but be in normal distribution, the sample of high ingredient and low component content is less, the sample Relatively centralized of intermediate amounts.If this
The indiscriminate foundation for directly participating in calibration model of a little samples, there may be regression forecasting results to tend to central value effect when prediction
(Dunneeffect).Common calibration set selection method has Sample set partitioning based on joint x-
Y distance (SPXY), Kennard Stone (KS) and Rank-Kennard Stone (Rank-KS).
Invention content
To solve the above-mentioned problems, the present invention provides a kind of biological products method for transferring near infrared model.The present invention adopts
With Rank-SPXY calibration samples collection division methods, on this basis, the PRS algorithms of binding model transfer realize different resolutions
Rate different wave length is counted the Model transfer of major-minor instrument, and the Partial Least Squares Regression (partial for quantitative forecast is established
Least squares regression, PLSR) model.The Model transfer strategy of Rank-SPXY-PRS is formd, and is successfully used
The assay research of total protein during biological products produce blood plasma alcohol precipitation.
To achieve the above object, the present invention uses following technical scheme：
One of the object of the invention provides a kind of biological products method for transferring near infrared model, includes the following steps：
(1) total protein is original during the blood plasma alcohol precipitation in main instrument and auxiliary instrument acquisition biological blood production of articles respectively
The NIR spectra of sample；
(2) selection of collection is corrected to the sample spectrum obtained using Rank-SPXY methods, wherein calibration set is used for
The foundation of quantitative model, predictive ability of the verification collection for verifying model；
(3) Model transfer is carried out to the FT-NIR spectrum of acquisition using PRS algorithms, FT-NIR spectrum is translated into together
The spectrum in the 1700 co-wavelength sections MicroNIR, and it is fitted comparison with the original spectrum of Micro NIR 1700；
(4) Pretreated spectra is carried out；
(5) PLSR models are established；
(6) prediction effect after being shifted near infrared correction carries out evaluation analysis.
The present invention combines PRS algorithms to carry out spectrum transfer by selecting the calibration samples for participating in Model transfer,
Keep the model of foundation more reliable and more stable.Calibration set is divided using Rank-SPXY methods with verification to collect, and existing method KS,
SPXY is compared with Rank-KS, and the sample set spatial distribution that this method divides is more reasonable, and utilizes the PLSR model evaluations established
Parameter also demonstrates the superiority of this method.The present invention inquires into the spectral window width of Model transfer PRS algorithms, finds
Selection window width is 7.The present invention is by comparing first derivative (FD), second dervative (SD), smooth (Smoothing), standard
(SNV), multiplicative scatter correction (MSC) and standardization (autoscale) is normalized to use to Pretreated spectra light alone or in combination
The influence of PLSR model performance parameters after spectrum transfer, finds calibration samples division methods+first derivative using Rank-SPXY
15 points of+SG is smooth+standardization combined method pretreatment after the model prediction ability established it is preferable.The present invention is by Model transfer
PLSR of the PLSR models afterwards with the PLSR models and original 1700 establishment of spectrum of Micro NIR of original FT-NIR establishment of spectrum
Model is compared, and the quality of spectrum and the predictive ability of model all increase after Model transfer.
The two of the object of the invention, provide a kind of NIR Spectroscopy Analysis Model, and the model is built up by above method.
The three of the object of the invention provide blood plasma alcohol of the NIR Spectroscopy Analysis Model in biological blood production of articles
Application during total protein content measures during heavy.
Beneficial effects of the present invention are：
A kind of biological products method for transferring near infrared model of the present invention is divided calibration set and is tested using Rank-SPXY methods
Card collection, the calibration samples selected participate in the Model transfer in later stage so that the PLSR model prediction abilities finally obtained obtain
It improves；One of important committed step as Model transfer, compared with traditional calibration samples division methods, this method obtains more
Effective calibration samples, sample space are added to be evenly distributed, and the near-infrared model predictive ability established is stronger.Simultaneously with PRS groups
It closes the Model transfer strategy formed and also shortens the quality inspection time, save manpower financial capacity's material resources.The foundation of this method contributes to model to turn
Move biological blood article manufacturing process parameter monitoring application, ensure blood product batch between consistency, improve the peace of drug
Full property and validity.
Description of the drawings
The accompanying drawings which form a part of this application are used for providing further understanding of the present application, and the application's shows
Meaning property embodiment and its explanation do not constitute the improper restriction to the application for explaining the application.
Fig. 1 is the NIR spectra (A of collection sample during blood plasma alcohol precipitation：a：The original spectrum of FT-NIR collecting samples, b：
The original spectrum of 1700 collecting samples of Micro NIR；B：Spectrum comparison under unified abscissa (wavelength) parameter, c:FT-NIR
Spectrum, d:1700 spectrum of Micro-NIR)；
Fig. 2 is that (a is compared in the sample PCA space distribution of four kinds of division calibration samples methods:SPXY；b:KS；c:Rank-
SPXY；d:Rank-KS.Stain and ash point, which respectively represent calibration set and verify, to be collected)；
Fig. 3 is PRS Model transfers algorithm used in the present invention and basis PDS algorithm principles and compares；
Fig. 4 is that gained spectrum acquires light with original Micro-NIR 1700 after carrying out spectrum transfer to FT-NIR in the present invention
The comparison of spectrum；
Total protein content PLSR prediction models during best blood plasma alcohol precipitation after the Model transfer that Fig. 5 obtains for the present invention
Figure.
Specific implementation mode
It is noted that following detailed description is all illustrative, it is intended to provide further instruction to the application.Unless another
It indicates, all technical and scientific terms used herein has usual with the application person of an ordinary skill in the technical field
The identical meanings of understanding.
It should be noted that term used herein above is merely to describe specific implementation mode, and be not intended to restricted root
According to the illustrative embodiments of the application.As used herein, unless the context clearly indicates otherwise, otherwise singulative
It is also intended to include plural form, additionally, it should be understood that, when in the present specification using term "comprising" and/or " packet
Include " when, indicate existing characteristics, step, operation, device, component and/or combination thereof.
As background technology is introduced, in the prior art common Model transfer sample set division methods need to be innovated and
It improves, to establish more preferably Model transfer PLSR near-infrareds prediction model, in order to solve technical problem as above, the application carries
A kind of more effective biological products method for transferring near infrared model is gone out, has included the following steps：
(1) total protein is original during the blood plasma alcohol precipitation in main instrument and auxiliary instrument acquisition biological blood production of articles respectively
The NIR spectra of sample；
(2) selection of collection is corrected to the sample spectrum obtained using Rank-SPXY methods, wherein calibration set is used for
The foundation of quantitative model, predictive ability of the verification collection for verifying model；
(3) Model transfer is carried out to the FT-NIR spectrum of acquisition using PRS algorithms, FT-NIR spectrum is translated into together
The spectrum in the 1700 co-wavelength sections MicroNIR, and it is fitted comparison with the original spectrum of Micro NIR 1700；
(4) Pretreated spectra is carried out；
(5) PLSR models are established；
(6) prediction effect after being shifted near infrared correction carries out evaluation analysis.
Preferably, near infrared spectra collection parameter is in step (1)：
All main instrument spectral data carry out original near infrared light by the transmission module of Antaris II FT-NIR spectrometers
Sample is loaded into the glass tubule that internal diameter is 3-5mm by the acquisition of spectrum；The scanning range of spectrum is 10000-4000cm-1, point
Resolution is 6-10cm-1, scanning times are 30-34 times, using air as reference, acquire a background per hour.
Preferably, near infrared spectra collection parameter is in step (1)：
All main instrument spectral data carry out original near infrared light by the transmission module of Antaris II FT-NIR spectrometers
Sample is loaded into the glass tubule that internal diameter is 4mm by the acquisition of spectrum；The scanning range of spectrum is 10000-4000cm-1, differentiate
Rate is 8cm-1, scanning times are 32 times, using air as reference, acquire a background per hour.
Preferably, near infrared spectra collection parameter is in step (1)：
All auxiliary instrument spectral data are collected by 1700 miniature near infrared spectrometers of Micro NIR；Spectral scan
Range 780-1650nm, light path 1mm are reference, time of integration 30000us with empty cuvette.
Preferably, step (2) Rank-SPXY methods are：Sample is ranked up by dependent variable number, then by sample etc.
It is divided into m parts；After selecting sample, SPXY formula is recycled to carry out subsequent correction samples selection；Wherein calibration set number of samples with test
The number of samples ratio of card collection is 2:1.
Preferably, the window width for analyzing to obtain Model transfer in step (3) by using Pearson correlation coefficient rij is
7-9。
Preferably, the window width for analyzing to obtain Model transfer in step (3) by using Pearson correlation coefficient rij is
7。
Preferably, step (4) preprocessing procedures are：By first derivative, 15 points of SG is smooth and standardization group share in
Pretreated spectra.
The two of the object of the invention, provide a kind of NIR Spectroscopy Analysis Model, and the model is built up by the above method.
The three of the object of the invention provide blood plasma alcohol precipitation mistake of the NIR Spectroscopy Analysis Model in biological blood production of articles
Application during total protein content measures in journey.
In order to enable those skilled in the art can clearly understand the technical solution of the application, below with reference to tool
The technical solution of the application is described in detail in the embodiment of body.
1 blood plasma alcohol precipitation process experiment and sample total protein content reference values
It is 5.95 ± 0.05 that raw blood plasma, which adds acetate buffer solution under the conditions of 4 DEG C and adjusts blood plasma pH, weighs blood plasma at this time
Solution quality, the 95% ethyl alcohol volume that should be added according to Mass Calculation.Blood plasma alcohol precipitation mistake is then carried out in low-temp reaction instrument
Journey, temperature setting is consistent with actual production reaction temperature, is -4.5 DEG C.Each batch takes 100mL plasma supernatants to be placed in 250mL's
In round-bottomed flask, when temperature is 0 DEG C start that ethyl alcohol, the speed that absolute ethyl alcohol is added in constant flow pump is added to be 0.875mL/min.Alcohol precipitation is opened
The measurement of protein content is used for before beginning and during alcohol precipitation every 2min samplings 0.3mL.The alcohol precipitation time of one batch is
21 samples are obtained in 40min, carry out the alcohol precipitation process of 8 normal batches altogether, there are 168 samples.Gained sample carries out
It centrifuges (6000rpm*15min), its supernatant is taken to carry out the analysis of total protein content.Total protein content assay method is biuret
Method, all sample standard deviation trust moneys domain medical test center carry out the measurement of related content, and the instrument used is Beckman AU 5800
Type automatic clinical chemistry analyzer.
The NIR spectra of 2 main instruments and auxiliary instrument acquires
All main instrument spectral data use the Transmission Modes of Antaris II FT-NIR spectrometers (hereinafter FT-NIR)
Block carries out the acquisition of original near infrared spectrum, sample is loaded into the glass tubule that internal diameter is 4mm.The scanning range of spectrum is
10000-4000cm-1, resolution ratio 8cm-1, scanning times are 32 times, using air as reference, acquire a background per hour.
Obtain (a) spectrogram that main instrument spectral is shown in Figure 1A.
All auxiliary instrument spectral data are collected by 1700 miniature near infrared spectrometers of Micro NIR.Spectrum is swept
Range 780-1650nm, light path 1mm are retouched, is reference, time of integration 30000us with empty cuvette.3 light of each sample collection
Spectrum, it is final spectrum to take average spectrum.Obtain (b) spectrogram that auxiliary instrument spectral is shown in Figure 1A.
It is that wavelength carries out spectrum comparison by the abscissa uniform units of two kinds of spectrum, sees Figure 1B spectrograms.Blood plasma raw material is through alcohol
I.e. protein, the flexible and deformation vibration of wherein C-H groups are main except the water removal of acquired solution main component, ethyl alcohol after heavy operation
In 3950-4400cm-1The two level frequency multiplication of (2273-2532nm), C-H stretching vibrations are located at 5700-5850cm-1(1709-
1754nm), the two level frequency multiplication of N-H stretching vibrations is located at 6500cm-1(1538nm), 6900cm-1(1449nm) is that the O-H of water stretches
The two level frequency multiplication of contracting and deformation vibration, 8100cm-1(1235nm) is the three-level frequency multiplication of C-H stretching vibrations.
The comparison of 3 calibration samples division methods
KS methods divide sample set principle be：The Euclidean distance of all samples between any two is calculated first, and chosen distance is most
Remote sample as calibration set, then calculate remaining sample between selected sample at a distance from, in the shortest distance it is relatively long away from
From be selected into verification collection, be specifically shown in formula (1).
SPXY is developed on the basis of KS methods, experiments have shown that SPXY methods, which can be efficiently used for NIR, quantifies mould
The foundation of type.It the advantage is that and take in the information of Y, it can be effective over multi-C vector space, so as to improve being built
The predictive ability of model.SPXY takes into account x variables and y variables when distance calculates between sample simultaneously, has added range formula
(2).SPXY methods d simultaneouslyxy(p, q) is instead of dx(p, q), while in order to ensure sample is in the having the same of the spaces x and y
Weight, by dx(p, q) and dy(p, q) respectively divided by their maximum values in data set, therefore standardized dxy(p, q) away from
It is (3) from formula：
Rank is data to be sized, and be divided into m sections, and certain n sample is chosen from each section as school
Positive collection, it is ensured that the uniformity that sample set is chosen, reference formula (4,5)
Rei∈[ymin+(i-1)*D,ymin+i*D],i∈[1,m] (5)
Rank and SPXY is combined, i.e., m sections are divided to sample data, after selecting sample with certain proportion, recycles SPXY
Formula carries out subsequent correction samples selection.It is a kind of new trial using Rank-SPXY method choice sample set division methods, this
Text compares the result attempted this with conventional sample set division methods (KS methods, SPXY methods, Rank-KS methods), and verification should
The Optimality of calibration samples division methods.
Sample is with 2:1 ratio is divided.Table 1 is the sample set information that distinct methods select, and is as a result shown
The calibration set and verification collection of Rank-KS methods and Rank-SPXY selections have similar SD values and mean value difference is smaller, verifies the dense of collection
Degree range occupy in the concentration range of calibration set, as a result more excellent.But Rank-KS methods are tested due to not accounting for Y, selection
The concentration range of card collection is not fully in the concentration range of calibration set.For the sample set division methods for preferably going out best, to four
The sample that kind division methods obtain carries out sample PCA and is analyzed, and distribution results are as shown in Figure 2.In Fig. 2 (a) (b) (d)
In SPXY, KS and Rank-KS method sample set division result figure, only there is not verification comprising calibration set sample in the circle of black
It includes wherein, to illustrate that the selection result of this area sample collection is not optimal, not up to verification collection sample is uniformly distributed in collect sample
The requirement of calibration set.The result figure that Fig. 2 (c) Rank-SPXY method sample sets divide, it is seen that sample is more uniform is distributed in for verification collection
In calibration set sample.
The different sample set division methods parameters of table 1 compare
4 Model transfers
Model transfer PRS algorithms are proposed based on PDS methods, and Fig. 3 is shown in concrete principle and difference.PRS and PDS algorithms are all
It is a kind of common multivariate calibration model transmission method, its partial correction and changeable flow characteristic make it be better than other models turn
Algorithm is moved, the variation of absorbance intensity, wavelength points can be deviated simultaneously and spectral peak broadening is corrected.
Specially：First in i-th of wavelength points of conversion spectrum or so, one window (i-k, i+w) of extension, Z is enablediIt indicates
Convert the absorbance matrix of spectrum total k+w+1 wavelength points from i-k to i+w：
Zi=[As,i-k,As,i-k+1,…,As,i+w-1,As,i+w] (6)
Then by i-th of wavelength points absorbance A of target optical spectrumm,iWith ZiConstruct a multivariate regression models：
Am,i=Zibi+ei (7)
This equation is solved by PLSR methods.By all regression coefficient biIt is placed on the leading diagonal of transition matrix F, and will
Other elements are set to 0, obtain a Diagonal matrix F (assuming that having P wavelength points)：
F=diag (b1 T,b2 T,…,bi T,…,bp T) (8)
A in its calculation formula (6)i,kFor the corresponding absorbance value of i-th of sample spectrum, k-th of wavelength points.Diagonal matrix
F acts on transition matrix, and spectrum to be transformed is converted into the required spectrum to match with target optical spectrum.
Then according to FT-NIR spectral characteristics select a window width, to establish the spectroscopic data in window width with
The PLSR models of spectrum corresponding points in Micro NIR 1700.Window width is important one of parameter in PDS algorithms.Herein
It is Pearson correlation coefficient r to quote formula (9)ij, according to i, the correlation between j determines best window width.xiFor main instrument
The average value of i-th of sample spectrum y-axis, yiSupplemented by i-th of sample spectrum y-axis of instrument average value, k=1,2 ..., m, m is wave
Long points.R values illustrate more close between i and j closer to 1, and fitting result is better, and window width at this time is best.
Influence of the odd number window width to Model transfer result between 7-21 has been investigated in experiment.Table 2 is to before and after transfer
The front and back related coefficient of each corresponding wave number conversion in 950-1650nm all bands section is for statistical analysis, as a result, it has been found that, with
The increase of window width, related coefficient constantly declines, and mean square deviation and variance constantly increase, therefore, final choice window width
It is 7.
Main instrument spectral translates into the Transfer Spectroscopy that range of wavelengths is 950-1650nm and sees Fig. 4.To the light after Model transfer
Spectrum is fitted operation with original 1700 spectrum of Micro NIR, finds the original figure spectrum and Model transfer of Micro NIR 1700
NIR light spectrogram afterwards, the two fit solution are preferable.And two kinds of collection of illustrative plates are compared, the fluctuation difference of the sample collection of illustrative plates after Model transfer
Obviously original figure spectrum is small between the sample than Micro NIR 1700.
The related coefficient statistical value of 2 all band of table (950-1650nm) different windows width number
5 preprocess methods select
This investigation and comparison derivative (FD and SD) combines the spectrum such as smooth (Smoothing), SNV, MSC and autoscale
Preprocess method, while combination correction sample division methods establish PLSR models together, have investigated the new samples collection proposed and have divided
The model prediction ability of method and commonsense method compares.Table 3 is that sample set division methods and preprocess method compositional modeling are predicted
Parameter it was found that, is Rank- with sample division methods by model parameter using RMSEP as the main evaluation parameter of model
SPXY methods, pre-process for 15 points of first derivative+SG it is smooth+standardization when model it is optimal.
The different calibration samples division methods of table 3 and the comparison of different pretreatments method PLSR modeling parameters
6PLSR modeling results compare
Table 4 compares for the PLSR model parameters established after main instrument, auxiliary instrument and Model transfer, and corrected sample divides
After Pretreated spectra, the R of modelc 2And Rp 2Modeling significantly improves when compared with without processing, and RMSEP is decreased obviously, after Model transfer
PLSR models establish moral PLSR model predictions ability than original main instrument and auxiliary instrument and all increase.Comprehensive entire experiment, says
The Rank-SPXY calibration samples division methods of bright proposition are more effective, while the Rank-SPXY-PRS Model transfer strategies combined can
To improve the validity of model.The LVs=for the best PLSR models that total protein content measures after the transfer of blood plasma alcohol precipitation process model
1, Rp 2=0.8395, RMSEP=0.934g/L, PLSR prediction model figure are shown in Fig. 5.
The PLSR model parameters established after 4. main instrument of table, auxiliary instrument and Model transfer compare
The foregoing is merely the preferred embodiments of the application, are not intended to limit this application, for the skill of this field
For art personnel, the application can have various modifications and variations.Within the spirit and principles of this application, any made by repair
Change, equivalent replacement, improvement etc., should be included within the protection domain of the application.
Claims (10)
1. a kind of biological products method for transferring near infrared model, which is characterized in that include the following steps：
(1) main instrument and auxiliary instrument acquire total protein original sample during the blood plasma alcohol precipitation in biological blood production of articles respectively
NIR spectra；
(2) selection of collection is corrected to the sample spectrum obtained using Rank-SPXY methods, wherein calibration set is for quantitative
The foundation of model, predictive ability of the verification collection for verifying model；
(3) Model transfer is carried out to the FT-NIR spectrum of acquisition using PRS algorithms, FT-NIR spectrum is translated into same Micro
The spectrum in the 1700 co-wavelength sections NIR, and it is fitted comparison with the original spectrum of Micro NIR 1700；
(4) Pretreated spectra is carried out；
(5) PLSR models are established；
(6) prediction effect after being shifted near infrared correction carries out evaluation analysis.
2. the method as described in claim 1, which is characterized in that near infrared spectra collection parameter is in step (1)：
All main instrument spectral data carry out original near infrared spectrum by the transmission module of Antaris II FT-NIR spectrometers
Sample is loaded into the glass tubule that internal diameter is 3-5mm by acquisition；The scanning range of spectrum is 10000-4000cm-1, resolution ratio
For 6-10cm-1, scanning times are 30-34 times, using air as reference, acquire a background per hour.
3. the method as described in claim 1, which is characterized in that near infrared spectra collection parameter is in step (1)：
All main instrument spectral data carry out original near infrared spectrum by the transmission module of Antaris II FT-NIR spectrometers
Sample is loaded into the glass tubule that internal diameter is 4mm by acquisition；The scanning range of spectrum is 10000-4000cm-1, resolution ratio is
8cm-1, scanning times are 32 times, using air as reference, acquire a background per hour.
4. the method as described in claim 1, which is characterized in that near infrared spectra collection parameter is in step (1)：
All auxiliary instrument spectral data are collected by 1700 miniature near infrared spectrometers of Micro NIR；Spectral scanning range
780-1650nm, light path 1mm are reference, time of integration 30000us with empty cuvette.
5. the method as described in claim 1, which is characterized in that step (2) Rank-SPXY methods are：Sample is pressed into dependent variable
Number is ranked up, and sample is then divided into m parts；After selecting sample, SPXY formula is recycled to carry out subsequent correction sample choosing
It selects；The number of samples ratio of wherein calibration set number of samples and verification collection is 2:1.
6. the method as described in claim 1, characterized in that step is analyzed in (3) by using Pearson correlation coefficient rij
Window width to Model transfer is 7-9.
7. the method as described in claim 1, characterized in that step is analyzed in (3) by using Pearson correlation coefficient rij
Window width to Model transfer is 7.
8. the method as described in claim 1, characterized in that step (4) preprocessing procedures are：By first derivative, SG 15
Point is smooth and standardization group is shared in Pretreated spectra.
9. a kind of NIR Spectroscopy Analysis Model, which is characterized in that built up by claim 1-5 the methods.
10. during blood plasma alcohol precipitation of the NIR Spectroscopy Analysis Model as claimed in claim 9 in biological blood production of articles
Application in total protein content measurement.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810339983.XA CN108548794A (en) | 2018-04-16 | 2018-04-16 | A kind of biological products method for transferring near infrared model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810339983.XA CN108548794A (en) | 2018-04-16 | 2018-04-16 | A kind of biological products method for transferring near infrared model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108548794A true CN108548794A (en) | 2018-09-18 |
Family
ID=63515170
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810339983.XA Pending CN108548794A (en) | 2018-04-16 | 2018-04-16 | A kind of biological products method for transferring near infrared model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108548794A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110736707A (en) * | 2019-09-16 | 2020-01-31 | 浙江大学 | Spectrum detection optimization method for spectrum model transfer from master instruments to slave instruments |
| CN111077102A (en) * | 2018-10-19 | 2020-04-28 | 中国石油化工股份有限公司 | Crude oil near infrared spectrum conversion and crude oil identification method measured by instruments of different models |
| CN112903625A (en) * | 2021-01-25 | 2021-06-04 | 北京工业大学 | Integrated parameter optimization modeling method for analyzing content of active substances in drugs based on partial least square method |
| CN115952402A (en) * | 2022-09-29 | 2023-04-11 | 南京林业大学 | Near-infrared model transfer standard sample set selection method based on binary dragonfly algorithm |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101832922A (en) * | 2010-05-19 | 2010-09-15 | 中国农业大学 | Method for transferring near infrared model of organic fertilizer product |
| CN104048938A (en) * | 2013-03-15 | 2014-09-17 | 山东大学 | Method for rapid determination of immunoglobulin G content in raw material plasma alcohol precipitation process |
| CN105628645A (en) * | 2015-12-28 | 2016-06-01 | 贵州中烟工业有限责任公司 | Infrared analysis model transfer method |
| CN106845548A (en) * | 2017-01-24 | 2017-06-13 | 清华大学 | Model transfer method based on high-spectral data |
-
2018
- 2018-04-16 CN CN201810339983.XA patent/CN108548794A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101832922A (en) * | 2010-05-19 | 2010-09-15 | 中国农业大学 | Method for transferring near infrared model of organic fertilizer product |
| CN104048938A (en) * | 2013-03-15 | 2014-09-17 | 山东大学 | Method for rapid determination of immunoglobulin G content in raw material plasma alcohol precipitation process |
| CN105628645A (en) * | 2015-12-28 | 2016-06-01 | 贵州中烟工业有限责任公司 | Infrared analysis model transfer method |
| CN106845548A (en) * | 2017-01-24 | 2017-06-13 | 清华大学 | Model transfer method based on high-spectral data |
Non-Patent Citations (3)
| Title |
|---|
| EVA-LOTTA BERGMAN等: "Transfer of NIR calibrations for pharmaceutical formulations between different instruments", 《JOURNAL OF PHARMACEUTICAL AND BIOMEDICAL ANALYSIS》 * |
| 张学博 等: "近红外多元校正模型传递的进展", 《药物分析杂质》 * |
| 李路 等: "单品种淡水活鱼数量的被动声学估计", 《农业机械学报》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111077102A (en) * | 2018-10-19 | 2020-04-28 | 中国石油化工股份有限公司 | Crude oil near infrared spectrum conversion and crude oil identification method measured by instruments of different models |
| CN111077102B (en) * | 2018-10-19 | 2022-06-24 | 中国石油化工股份有限公司 | Crude oil near infrared spectrum conversion and crude oil identification method measured by instruments of different models |
| CN110736707A (en) * | 2019-09-16 | 2020-01-31 | 浙江大学 | Spectrum detection optimization method for spectrum model transfer from master instruments to slave instruments |
| CN112903625A (en) * | 2021-01-25 | 2021-06-04 | 北京工业大学 | Integrated parameter optimization modeling method for analyzing content of active substances in drugs based on partial least square method |
| CN112903625B (en) * | 2021-01-25 | 2024-01-19 | 北京工业大学 | Integrated parameter optimization modeling method for analyzing content of active substances in medicine based on partial least square method |
| CN115952402A (en) * | 2022-09-29 | 2023-04-11 | 南京林业大学 | Near-infrared model transfer standard sample set selection method based on binary dragonfly algorithm |
| CN115952402B (en) * | 2022-09-29 | 2023-06-27 | 南京林业大学 | Method for selecting near-infrared model transfer standard sample set based on binary dragonfly algorithm |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108548794A (en) | A kind of biological products method for transferring near infrared model | |
| CN104062257B (en) | A kind of based on the method for general flavone content near infrared ray solution | |
| US20070211247A1 (en) | Visible/Near-Infrared Spectrometry And Its Device | |
| CN102590129B (en) | Method for detecting content of amino acid in peanuts by near infrared method | |
| CN109799207A (en) | The quantitative detecting method of talcum powder is mixed in the root of Dahurain angelica based on near-infrared spectrum analysis | |
| CN104897607A (en) | Food modeling and rapid detecting integration method and system adopting portable NIRS (near infrared spectroscopy) | |
| CN104062258B (en) | Method for rapid determination of soluble solids in compound ass-hide glue pulp by near infrared spectroscopy | |
| CN101713731A (en) | Method for distinguishing coating quality of medicine preparation | |
| WO2008004665A1 (en) | Method of testing, and apparatus therefor, as to cancer, systemic lupus erythematosus (sle) or antiphospholipid antibody syndrome, using near-infrared ray | |
| Dubois et al. | Peer reviewed: IR spectroscopy in clinical and diagnostic applications | |
| CN103592255A (en) | Soft method for measuring total protein content of donkey-hide gelatin skin solution on basis of near infrared spectrum technology | |
| CN104062259A (en) | Method for rapid determination of total saponin content in compound ass-hide glue pulp by near infrared spectroscopy | |
| CN109765214A (en) | The detection method of infertile patients serum based on Surface enhanced Raman spectroscopy | |
| CN108982406A (en) | A kind of soil nitrogen near-infrared spectral characteristic band choosing method based on algorithm fusion | |
| CN107917897A (en) | The method of the special doctor's food multicomponent content of near infrared ray | |
| CN108489929A (en) | Ginseng, Radix Notoginseng and the legal base source Panax polysaccharide of three kinds of American Ginseng discrimination method | |
| CN105717066A (en) | Near-infrared spectrum recognition model based on weighting association coefficients | |
| CN107402192A (en) | A kind of method of quick analysis essence and flavoring agent quality stability | |
| Guilment et al. | Determination of polybutadiene microstructures and styrene–butadiene copolymers composition by vibrational techniques combined with chemometric treatment | |
| CN107462524A (en) | The a variety of property quick analytic instruments of pitch | |
| CN110186871A (en) | A kind of method of discrimination in the fresh tea leaves place of production | |
| Van den Broek et al. | Application of a spectroscopic infrared focal plane array sensor for on-line identification of plastic waste | |
| CN109540837A (en) | The method that near-infrared quickly detects Boehmeria nivea leaves wood fibre cellulose content | |
| CN103760130A (en) | Method for measuring tween-80 content in compound musk injection by using near-infrared spectrum | |
| Yang et al. | Feasibility of terahertz spectroscopy for hybrid purity verification of rice seeds |
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 |