CN109632696A - A kind of inexpensive near-infrared spectrum method identifying medicinal tablet source - Google Patents
A kind of inexpensive near-infrared spectrum method identifying medicinal tablet source Download PDFInfo
- Publication number
- CN109632696A CN109632696A CN201811524070.1A CN201811524070A CN109632696A CN 109632696 A CN109632696 A CN 109632696A CN 201811524070 A CN201811524070 A CN 201811524070A CN 109632696 A CN109632696 A CN 109632696A
- Authority
- CN
- China
- Prior art keywords
- infrared
- spectrum
- vector
- matrix
- infrared spectrum
- 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
- 238000002329 infrared spectrum Methods 0.000 title claims abstract description 55
- 238000000034 method Methods 0.000 title claims abstract description 54
- 238000001228 spectrum Methods 0.000 claims abstract description 62
- 239000013598 vector Substances 0.000 claims abstract description 62
- 238000004458 analytical method Methods 0.000 claims abstract description 57
- 238000010239 partial least squares discriminant analysis Methods 0.000 claims abstract description 8
- 239000011159 matrix material Substances 0.000 claims description 48
- 230000004044 response Effects 0.000 claims description 13
- 238000012360 testing method Methods 0.000 claims description 11
- 238000012549 training Methods 0.000 claims description 11
- 230000003595 spectral effect Effects 0.000 claims description 9
- 238000000354 decomposition reaction Methods 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 6
- 239000013307 optical fiber Substances 0.000 claims description 5
- 238000005286 illumination Methods 0.000 claims description 4
- 230000010354 integration Effects 0.000 claims description 4
- 229910000530 Gallium indium arsenide Inorganic materials 0.000 claims description 3
- 238000012937 correction Methods 0.000 claims description 3
- 238000002790 cross-validation Methods 0.000 claims description 3
- 238000011068 loading method Methods 0.000 claims description 3
- 239000011248 coating agent Substances 0.000 claims description 2
- 238000000576 coating method Methods 0.000 claims description 2
- 239000003814 drug Substances 0.000 abstract description 18
- 238000005516 engineering process Methods 0.000 abstract description 8
- 239000003826 tablet Substances 0.000 description 33
- 229940079593 drug Drugs 0.000 description 14
- 239000003962 counterfeit drug Substances 0.000 description 8
- 238000001514 detection method Methods 0.000 description 4
- 238000011161 development Methods 0.000 description 4
- 239000007787 solid Substances 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 239000003795 chemical substances by application Substances 0.000 description 3
- 238000004587 chromatography analysis Methods 0.000 description 3
- 230000036541 health Effects 0.000 description 3
- 239000002245 particle Substances 0.000 description 3
- 238000010183 spectrum analysis Methods 0.000 description 3
- MQTOSJVFKKJCRP-BICOPXKESA-N azithromycin Chemical compound O([C@@H]1[C@@H](C)C(=O)O[C@@H]([C@@]([C@H](O)[C@@H](C)N(C)C[C@H](C)C[C@@](C)(O)[C@H](O[C@H]2[C@@H]([C@H](C[C@@H](C)O2)N(C)C)O)[C@H]1C)(C)O)CC)[C@H]1C[C@@](C)(OC)[C@@H](O)[C@H](C)O1 MQTOSJVFKKJCRP-BICOPXKESA-N 0.000 description 2
- 239000000470 constituent Substances 0.000 description 2
- 230000003247 decreasing effect Effects 0.000 description 2
- 230000007812 deficiency Effects 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000001819 mass spectrum Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 238000005220 pharmaceutical analysis Methods 0.000 description 2
- 238000000513 principal component analysis Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 239000000243 solution Substances 0.000 description 2
- 206010013786 Dry skin Diseases 0.000 description 1
- 239000003708 ampul Substances 0.000 description 1
- 229960004099 azithromycin Drugs 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000003115 biocidal effect Effects 0.000 description 1
- 238000005251 capillar electrophoresis Methods 0.000 description 1
- 239000003153 chemical reaction reagent Substances 0.000 description 1
- 238000009500 colour coating Methods 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 239000013078 crystal Substances 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 238000003255 drug test Methods 0.000 description 1
- 238000001035 drying Methods 0.000 description 1
- 238000005469 granulation Methods 0.000 description 1
- 230000003179 granulation Effects 0.000 description 1
- 238000004128 high performance liquid chromatography Methods 0.000 description 1
- -1 homogeneity Substances 0.000 description 1
- 238000011065 in-situ storage Methods 0.000 description 1
- 239000004615 ingredient Substances 0.000 description 1
- 238000002347 injection Methods 0.000 description 1
- 239000007924 injection Substances 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 239000003120 macrolide antibiotic agent Substances 0.000 description 1
- 229940041033 macrolides Drugs 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000002156 mixing Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010905 molecular spectroscopy Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000007935 oral tablet Substances 0.000 description 1
- 229940096978 oral tablet Drugs 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000010422 painting Methods 0.000 description 1
- 239000008194 pharmaceutical composition Substances 0.000 description 1
- 230000000704 physical effect Effects 0.000 description 1
- 239000000843 powder Substances 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 231100000331 toxic Toxicity 0.000 description 1
- 230000002588 toxic effect Effects 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
Abstract
The invention discloses a kind of inexpensive near-infrared spectrum methods for identifying medicinal tablet source, comprising: constructs near-infrared analysis system based on small-sized integrating sphere and Minitype infrared spectrometer；The near infrared spectrum of background spectrum and several different manufacturers tablet samples of the same name is acquired, and variable standardization pretreatment is carried out to the near infrared spectrum；Near-infrared analysis model is constructed based on partial least squares discriminant analysis algorithm, obtains the coefficient vector of near-infrared analysis model；The near infrared spectrum of the unknown medicinal tablet sample of acquisition is projected in the coefficient vector of the near-infrared analysis model, identifies the unknown medicinal tablet.Compared with conventional medicament identification technology, the present invention is at low cost, analysis mode is simple, quickly, has stronger practicability and generalization.
Description
Technical field
The invention belongs to the technical fields of Pharmaceutical Analysis test, and in particular to a kind of low cost for identifying medicinal tablet source
Near-infrared spectrum method.
Background technique
Adulterated drug not only threatens the health of the people, also constrains the fast development of social economy, has become generation
Various countries, boundary especially developing country's facing the problem of, or even be considered being only second to the second largest public hazards of drugs.According to generation
Boundary's health organization and media investigation, global counterfeit drug annual sales amount have broken through 40,000,000,000 dollars, every year also in the speed increasing with about 13%
It is long.In China, counterfeit drug problem is also very prominent, has seriously endangered the sound development in medical industry and market, has hit fake and inferior medicine
Product task abnormity is urgent.However, today, with the development of science and technology, traditional counterfeit drug, comprising pretending to be drug, with non-drug with low value drug
Pretend to be high price drug, it is fewer and fewer, instead increased or decreased in Medicine prescription privately without approval activity at
Divide, pretend to be brand name drug etc. with the drug that ordinary enterprises produce.It has also been found that the case for thering is true medicine and counterfeit drug to load in mixture in practice, to disease
The threat of people's health is more hidden, huger.
Currently, counterfeit drug problem emerges one after another, and counterfeit drug form is varied, and false making means mesh benefit is brilliant, and tends to " Gao Shui
Flat, high iq " development, proposes very big challenge for analysis and detection technology.Counterfeit drug identifies generally without fixed mode.For more
Complicated counterfeit drug generally requires mutually to prove in laboratory integration with a variety of qualitative and quantitative means, this traditional methods consumption
Duration uses toxic reagent, it may be necessary to large-scale instrument and those skilled in the art.However, an ideal analysis detection skill
Art should have the characteristics that rapidity, simplicity, economy, environmental-friendly.It is short of any one, can all limit the analysis skill
The popularization and use of art.It sums up, the analytical technology for adulterated drug mainly has: chemical method, chromatography, spectroscopic methodology.Chemistry
Extraction process in method can make troubles to live high-flux fast screening.Chromatography is always the master of such analysis detection task
Stream method, such as high performance liquid chromatography (HPLC), high performance capillary electrophoresis (HPCE) etc., these chromatographic techniques usually with second level battle array
The combination of the detecting instruments such as column detector (DAD), mass spectrum (MS), instrument cost is high, time-consuming, troublesome maintenance.Spectroscopic methodology is especially
Molecular Spectroscopy have it is fast it is simple, quick, sensitive, without sample treatment, can the unique advantages such as in situ detection, feature rich, tool
There is natural advantage.Spectrum analysis all have it is common be characterized in, the basic information for analyzing foundation is all that microcosmic particle moves
The signal of feature, analytic process load-bearing analysis information is all the spectrum of sample to be tested.Spectrum analysis is the Spectral Properties according to sample
Sign, by spectrum parameter and it is to be measured between mathematical model come realize analysis.Wave-length coverage is in the close red of 770-2500nm
External spectrum causes extensive attention in recent years, and the analysis information of the spectrum area carrier is mainly hydric group (C-H, O- in molecule
H, N-H, S-H, P-H etc.) vibration frequency multiplication and sum of fundamental frequencies characteristic information, therefore signal is weaker, can be used for lossless and on-line analysis.
However, instrument record is usually the apparent spectral comprising real spectrum, to spectrum due to the influence and uncertainty of background
Using making troubles.Can generally, the multiple information of near infrared spectrum carrying has complicated, overlapping, the feature changed, closely
Infrared analysis is that weak information to be measured is extracted from apparent spectral under complicated, overlapping, variation background, this is also near infrared light
The root and key of spectrum analysis difficulty, this needs the support of chemometric techniques.
Tablet medicine is the most commonly used a kind of solid orally ingestible of purposes, compared with powder ampoule agent for injection etc., oral tablet
Agent drug usually contains more auxiliary material, and usually requires in production process by kinds of processes mistakes such as mixing, granulation, dryings
Journey；The similar product of different enterprises, due to prescription difference, the composition of different auxiliary material can be entirely different, the content of active constituent
Also different, these result in the difference of the near infrared spectrum of the preparation with the identical name of an article, containing identical active constituent.This
Outside, the physical property of pharmaceutical formulation ingredient, such as granular size, distribution of particles, homogeneity, crystal form also have spectrum biggish
It influences.
Although near-infrared spectrum technique is used widely in terms of Pharmaceutical Analysis, it is close to be generally based on large-scale Fourier
Infrared spectrometer carries out spectra collection and analysis, at high cost, is not easy to promote and apply in base.
Summary of the invention
It is an object of the invention to be directed to above-mentioned deficiency in the prior art, provide it is a kind of identification medicinal tablet source it is low
Cost near-infrared spectrum method, to solve the above problems.
In order to achieve the above objectives, the technical solution adopted by the present invention is that:
A kind of inexpensive near-infrared spectrum method identifying medicinal tablet source comprising:
Near-infrared analysis system is constructed based on small-sized integrating sphere and Minitype infrared spectrometer；
The near infrared spectrum of background spectrum and several different manufacturers tablet samples of the same name is acquired, and near infrared spectrum is carried out
Variable standardization pretreatment；
Near-infrared analysis model is constructed based on partial least squares discriminant analysis algorithm, obtains the coefficient of near-infrared analysis model
Vector；
By the near infrared spectrum of the unknown medicinal tablet sample of acquisition project to the coefficient of the near-infrared analysis model to
In amount, the unknown medicinal tablet is identified.
Preferably, the near infrared spectrum of every class tablet samples of acquisition is marked, obtains label vector Y；
Variable standardization pretreatment is carried out to the near infrared spectrum
Wherein, xjFor j-th of response of spectrum,For the average value of spectral response, j is the location number of spectral response value, p
The response points for including for spectrum.
Preferably, the near infrared spectrum of every class tablet samples of acquisition is randomly divided into two subsets, and one includes 2/3 spectrum
Subset as training set for constructing near-infrared analysis model；Another includes that the subset of 1/3 spectrum is used to survey as test set
Try near-infrared analysis model.
Preferably, it using Partial Least Squares, is decomposed while similar with concentration matrix Y progress to spectrum matrix X:
Wherein, score matrix T and R respectively represents the spectrum after eliminating most of interference and classification information, V, Q are respectively
T loading matrix corresponding with R, EXFor spectrum residual matrix, EYFor concentration residual matrix.
Preferably, the method for the coefficient vector of near-infrared analysis model is calculated using nonlinear iterative partial least square are as follows:
S1, any of note concentration matrix Y are classified as r, and using r as the primary iteration vector for decomposing X；
S2, T matrix decomposition vector t is replaced with r, calculates vT:
vT=rTX/(rTr)
Wherein, t is a column of T matrix, vTFor VTThe row of matrix, rTFor the transposed vector of r；
S3, to the vTIt is normalized:
vT newFor the vector after standardization, vT oldFor original vector；
S4, according to the vTVector t is calculated,
T=Xt/ (vTv)
Wherein, t is a column of T matrix；
S5, R matrix decomposition vector r is replaced with t and calculates qT
qT=tTY/(tTt)
Wherein, qTFor QTA line of matrix；
S6, to the qTIt is normalized
Wherein, qT newFor the vector after standardization, qT oldFor original vector；
S7, according to the qTCalculate vector r
R=Yq/ (qTq)
Wherein, r is a column of R matrix；
S8, judge whether t restrains, if do not restrained, continue iteration in return step S2；If convergence, establishes r, t it
Between relationship: b=rTt/(tTT), wherein b is coefficient vector；
S9, E is calculatedX=X-tvT, EY=Y-rqT=Y-btqT,
Then with EXInstead of X, with EYInstead of Y, repeats above procedure and calculate next factor, until obtaining the required factor
Number；
S10, for unknown medicinal tablet sample spectra XIt is unknown, utilize X=TVTAnd the V that correction course obtainsTIt is calculated
TIt is unknown, then by r, the relationship between t obtains RIt is unknown, finally according to Y=RQTObtain YIt is unknown；The calculating process of step S10 is integrated, is obtained
To a coefficient vector b or coefficient matrix B.
Preferably, the near infrared spectrum of the unknown medicinal tablet sample of acquisition is projected into the near-infrared analysis model
In coefficient vector, the method for the identification unknown medicinal tablet are as follows:
Unknown medicinal tablet sample spectra X is calculated according to y=Xb or Y=XBIt is unknown, wherein b is coefficient vector, B coefficient square
Battle array.
Preferably, near-infrared analysis factor of a model number is controlled using prediction residual square PRESS cooperation cross validation,
PRESS is defined as:
Wherein, n is training set sample number, f be because of subnumber,yijIt is the predicted value and reference value of sample label.
Preferably, near-infrared analysis system includes the small-sized integrating sphere and Minitype infrared spectrum connected by optical fiber.
Preferably, the white painting whether there is or not the uniform diffuse reflection of wavelength selectivity is applied in the inner surface of the small-sized integrating sphere
Expect, the illumination in ball on any direction is equal.
Preferably, spectra collection is by integrating sphere, InGaAs detector array, and the time of integration is set as 20ms, and spectrum is adopted
Integrate wavelength region as 900-1700nm, the average time of spectrum is 2.
The inexpensive near-infrared spectrum method in identification medicinal tablet source provided by the invention, has beneficial below
The present invention is based on the near-infrared analysis systems that optical fiber technology and small-sized near infrared spectrometer have built low cost, use
Small-sized integrating sphere establishes analysis model as sampling apparatus, by partial least squares discriminant analysis algorithm, actually a coefficient
Vector easily can carry out direct nondestructive analysis to solid tablet by projection of the sample spectra on coefficient vector.With biography
System drug identification technology is compared, and the present invention is at low cost, analysis mode is simple, quickly, has stronger practicability and generalization.
Except this, the present invention is high to drug test accuracy rate, by analysis of cases, is tested on training set and test set
Card, accuracy rate is up to 100%, it was demonstrated that the accuracy of the method for the present invention.
Small-sized integrating sphere acquires spectrum, applies that whether there is or not the white of the uniform diffuse reflection of wavelength selectivity in the inner surface of integrating sphere
Color coating, the illumination in ball on any direction are equal.Light source (sample is irreflexive) any point on ball wall generates
Illuminance is that the illuminance generated by multiple reflections light is formed by stacking, and since signal light passes through integrating sphere space integral, can be had
Effect overcomes the influence of enchancement factor in diffusing reflection measurement, greatly improves data stability and repeatability.
Detailed description of the invention
Fig. 1 is the inexpensive near-infrared spectrum method flow chart for identifying medicinal tablet source.
Fig. 2 is the close red of four producer's azithromycins of inexpensive near-infrared spectrum method in identification medicinal tablet source
External spectrum.
Fig. 3 is being averaged for four producer's azithromycins of inexpensive near-infrared spectrum method in identification medicinal tablet source
Near infrared spectrum.
Fig. 4 is the variance percentage for the inexpensive near-infrared spectrum method explanation for identifying medicinal tablet source with number of principal components
Variation.
Fig. 5 is to identify the inexpensive near-infrared spectrum method in medicinal tablet source based on the three-dimensional scatterplot of first three principal component
Figure.
Fig. 6 is the variance percentage for the inexpensive near-infrared spectrum method explanation for identifying medicinal tablet source with minimum two partially
Multiply the variation because of subnumber.
Fig. 7 is to identify the inexpensive near-infrared spectrum method analysis model in medicinal tablet source on training set and test set
Performance.
Fig. 8 be identify medicinal tablet source inexpensive near-infrared spectrum method offset minimum binary discrimination model coefficient to
Amount.
Specific embodiment
A specific embodiment of the invention is described below, in order to facilitate understanding by those skilled in the art this hair
It is bright, it should be apparent that the present invention is not limited to the ranges of specific embodiment, for those skilled in the art,
As long as various change is in the spirit and scope of the present invention that the attached claims limit and determine, these variations are aobvious and easy
See, all are using the innovation and creation of present inventive concept in the column of protection.
According to one embodiment of the application, with reference to Fig. 1, the inexpensive near-infrared in the identification medicinal tablet source of this programme
Spectrographic technique constructs near-infrared analysis system based on small-sized integrating sphere and Minitype infrared spectrometer；
The near infrared spectrum of background spectrum and several different manufacturers tablet samples of the same name is acquired, and near infrared spectrum is carried out
Variable standardization pretreatment；
Near-infrared analysis model is constructed based on partial least squares discriminant analysis algorithm, obtains the coefficient of near-infrared analysis model
Vector；
By the near infrared spectrum of the unknown medicinal tablet sample of acquisition project to the coefficient of the near-infrared analysis model to
In amount, the unknown medicinal tablet is identified.
The above method is described in detail below
Macrolides second generation antibiotic to acquire four producers --- for azithromycin.
Near-infrared analysis system includes small-sized integrating sphere, optical fiber and small-sized near infrared spectrometer, and small-sized integrating sphere passes through light
Fine and small-sized near infrared spectrometer.
Using small-sized integrating sphere acquire spectrum, integrating sphere inner surface apply whether there is or not the uniform diffuse reflections of wavelength selectivity
White coating, the illumination in ball on any direction is equal.Is produced from light source (sample is irreflexive) any point on ball wall
Raw illuminance is that the illuminance generated by multiple reflections light is formed by stacking, since signal light passes through integrating sphere space integral,
The influence that enchancement factor in diffusing reflection measurement can effectively be overcome, greatly improves data stability and repeatability.
Spectra collection is set as 20ms, spectra collection wavelength zone by integrating sphere, InGaAs detector array, the time of integration
Domain is 900-1700nm, and the average time of spectrum is 2.
Come respectively to every class sample in four different provinces, is marked using " 1,2,3,4 ", and constitute label vector Y.
The near infrared spectrum of several different manufacturers tablet samples of the same name of acquisition is subjected to variable standardization pretreatment, is used for
Prominent useful information, is removed garbage, the average value of the spectrum, then the standard deviation divided by the spectrum is subtracted using original spectrum
Difference:
Wherein, xjFor j-th of response of spectrum,For the average value of spectral response, j is the location number of spectral response value, p
The response points for including for spectrum.
Variable standardization can preferably cut down error brought by solid particle size, surface scattering and change in optical path length.
Every class spectrum of acquisition according to being divided to mode to be divided into two subsets at random, and a subset comprising 2/3 spectrum is as instruction
Practice collection and establish model, separately the subset just comprising 1/3 spectrum carrys out test model as test set.It can be by training set and test set
Upper progress drug identification verifying, to test the accuracy rate of the method for the present invention.
Near-infrared analysis model is constructed based on partial least squares discriminant analysis algorithm.Offset minimum binary rule is further,
It not only decomposes spectrum matrix X, and while label vector (response vector, categorization vector) Y also carries out similar point
Solution, it may be assumed that
Wherein, score matrix T and R respectively represents the spectrum after eliminating most of interference and classification information, V, Q are respectively
T loading matrix corresponding with R, EXFor spectrum residual matrix, EYFor concentration residual matrix.
Except this, the factor of Y is considered when decomposing X, vice versa；Exchanging when passing through iteration iterative vectorized makes two matrixes
Decomposable process is combined into one.
The method of the coefficient vector of near-infrared analysis model is calculated using nonlinear iterative partial least square are as follows:
S1, any of note concentration matrix Y are classified as r, and using r as the primary iteration vector for decomposing X；
S2, T matrix decomposition vector t is replaced with r, calculates vT:
vT=rTX/(rTr)
Wherein, t is a column of T matrix, vTFor VTThe row of matrix, rTFor the transposed vector of r；
S3, to the vTIt is normalized:
vT newFor the vector after standardization, vT oldFor original vector；
S4, according to the vTVector t is calculated,
T=Xt/ (vTv)
Wherein, t is a column of T matrix；
S5, R matrix decomposition vector r is replaced with t and calculates qT
qT=tTY/(tTt)
Wherein, qTFor QTA line of matrix；
S6, to the qTIt is normalized
Wherein, qT newFor the vector after standardization, qT oldFor original vector；
S7, according to the qTCalculate vector r
R=Yq/ (qTq)
Wherein, r is a column of R matrix；
S8, judge whether t restrains, i.e., | | tPrevious round-tIt is latter to swing| | whether less than one threshold value if do not restrained returns to step
Continue iteration in rapid S2, at this time the r obtained in step S7；Otherwise, r, the relationship between t: b=r are establishedTt/(tTt)；
S9, E is calculatedX=X-tvT, EY=Y-rqT=Y-btqT, then with EXInstead of X, with EYInstead of Y, above procedure is repeated
Next factor is calculated, it is required because of subnumber until obtaining；
S10, for unknown sample spectrum XIt is unknown, utilize X=TVTAnd the V that correction course obtainsTT can be obtainedIt is unknown, then by r,
Relationship between t obtains RIt is unknown, finally according to Y=RQTY can be obtainedIt is unknown。
In practice, generally the calculating process of the step S10 is integrated, finally obtains a coefficient vector b (one pack system
It is quantitative) or coefficient matrix B (multicomponent is quantitative), analysis task can be conveniently realized according to y=Xb or Y=XB.
By the near infrared spectrum of the unknown medicinal tablet sample of acquisition project to the coefficient of the near-infrared analysis model to
In amount, the method for the identification unknown medicinal tablet are as follows:
Unknown medicinal tablet sample spectra X is calculated according to y=Xb or Y=XBIt is unknown, wherein b is coefficient vector, B coefficient square
Battle array.
In the modeling of the above Partial Least Squares, the most key is how to determine to establish required for model because of subnumber,
Generally, with the increase because of subnumber, load vectors gradually decrease the significance level of modeling, to a certain extent after, load to
Amount will become the noise of model.If use because subnumber deficiency, may make model owe training, can not reflect spectrum comprehensively
The relationship of matrix and concentration matrix；Conversely, being then likely to occur over-fitting, i.e., model has perfection in analyzing and training sample
Performance, but performance is remarkably decreased when predicting unknown sample.
The present invention uses prediction residual square PRESS (Prediction Residual Error Sum of Squares)
Cooperation cross validation carrys out Controlling model because of subnumber, PRESS is defined as:
In formula, n is training set sample number, f be because of subnumber,yijIt is the predicted value and reference value of sample label (classification)
(actual value).It is noted that the possible non-integer of the predicted value of model, is rounded using rounding up.
It is the near infrared spectrum of four producer's azithromycins with reference to Fig. 2.
It is the average near infrared spectrum of four producer's azithromycins with reference to Fig. 3, although spectrum is very much like, but still
Can difference therein, illustrate the method for the present invention acquisition spectrum include four different manufacturers Zitromax of valuable qualitative recognition
The information of plain piece agent.
For the distribution trend of further analysis and observation data, principal component analysis (PCA) is carried out to spectrum matrix.Principal component point
Analysis is a kind of classical way being widely used, and can realize variable compression and feature extraction simultaneously, it will be original by linear projection
High dimensional input vector (including all wavelengths variable) is changed into the vector (dimension is number of principal components) of a low-dimensional, can describe
The Main change trend of data is again irrelevant.
With reference to Fig. 4, the variance percentage of explanation is given with the variation of number of principal components, it is seen that three Principal Component Explanations are big
In 98% population variance.
With reference to Fig. 5, the three-dimensional scatter plot based on first three principal component is given, as seen from the figure, three principal components are good
Divide the spectrum of different manufacturers.
It with reference to Fig. 6, is modeled using partial least squares discriminant analysis, the conjunction of 6 offset minimum binary factors is shown in figure
It is suitable and enough.
Performance of the least square discriminant analysis model on training set and test set is given with reference to Fig. 7, is reached
100%.
With reference to Fig. 8, the coefficient vector of offset minimum binary discrimination model is given, where it can be seen that mostly important section
Near 1400nm.
The present invention is based on the near-infrared analysis systems that optical fiber technology and small-sized near infrared spectrometer have built low cost, use
Small-sized integrating sphere establishes analysis model as sampling apparatus, by partial least squares discriminant analysis algorithm, actually a coefficient
Vector easily can carry out direct nondestructive analysis to solid tablet by projection of the sample spectra on coefficient vector.With biography
System drug identification technology is compared, and the present invention is at low cost, analysis mode is simple, quickly, has stronger practicability and generalization.
Although being described in detail in conjunction with specific embodiment of the attached drawing to invention, should not be construed as to this patent
Protection scope restriction.In range described by claims, those skilled in the art are without creative work
The various modifications and deformation made still belong to the protection scope of this patent.
Claims (10)
1. a kind of inexpensive near-infrared spectrum method for identifying medicinal tablet source characterized by comprising
Near-infrared analysis system is constructed based on small-sized integrating sphere and Minitype infrared spectrometer；
The near infrared spectrum of background spectrum and several different manufacturers tablet samples of the same name is acquired, and the near infrared spectrum is carried out
Variable standardization pretreatment；
Based on partial least squares discriminant analysis algorithm construct near-infrared analysis model, obtain near-infrared analysis model coefficient to
Amount；
The near infrared spectrum of the unknown medicinal tablet sample of acquisition is projected in the coefficient vector of the near-infrared analysis model,
Identify the unknown medicinal tablet.
2. the inexpensive near-infrared spectrum method in identification medicinal tablet source according to claim 1, it is characterised in that:
The near infrared spectrum of every class tablet samples of acquisition is marked, label vector Y is obtained；
Variable standardization pretreatment is carried out to the near infrared spectrum
Wherein, xjFor j-th of response of spectrum,For the average value of spectral response, j is the location number of spectral response value, and p is light
The response points that spectrum includes.
3. the inexpensive near-infrared spectrum method in identification medicinal tablet source according to claim 1, it is characterised in that: adopt
The near infrared spectrum of every class tablet samples of collection is randomly divided into two subsets, and a subset comprising 2/3 spectrum is as training set
For constructing near-infrared analysis model；Another includes that the subset of 1/3 spectrum is used to test near-infrared analysis model as test set.
4. the inexpensive near-infrared spectrum method in identification medicinal tablet source according to claim 1, it is characterised in that:
Using Partial Least Squares, decomposed while similar with concentration matrix Y progress to spectrum matrix X:
Wherein, score matrix T and R respectively represents the spectrum after eliminating most of interference and classification information, V, Q are T and R respectively
Corresponding loading matrix, EXFor spectrum residual matrix, EYFor concentration residual matrix.
5. the inexpensive near-infrared spectrum method in identification medicinal tablet source according to claim 1, which is characterized in that adopt
The method of the coefficient vector of near-infrared analysis model is calculated with nonlinear iterative partial least square are as follows:
S1, any of note concentration matrix Y are classified as r, and using r as the primary iteration vector for decomposing X；
S2, T matrix decomposition vector t is replaced with r, calculates vT:
vT=rTX/(rTr)
Wherein, t is a column of T matrix, vTFor VTThe row of matrix, rTFor the transposed vector of r；
S3, to the vTIt is normalized:
For standardization after vector,For original vector；
S4, according to the vTVector t is calculated,
T=Xt/ (vTv)
Wherein, t is a column of T matrix；
S5, R matrix decomposition vector r is replaced with t and calculates qT
qT=tTY/(tTt)
Wherein, qTFor QTA line of matrix；
S6, to the qTIt is normalized
Wherein,For standardization after vector,For original vector；
S7, according to the qTCalculate vector r
R=Yq/ (qTq)
Wherein, r is a column of R matrix；
S8, judge whether t restrains, if do not restrained, continue iteration in return step S2；If convergence, establishes r, between t
Relationship: b=rTt/(tTT), wherein b is coefficient vector；
S9, E is calculatedX=X-tvT, EY=Y-rqT=Y-btqT,
Then with EXInstead of X, with EYInstead of Y, repeats above procedure and calculate next factor, it is required because of subnumber until obtaining；
S10, for unknown medicinal tablet sample spectra XIt is unknown, utilize X=TVTAnd the V that correction course obtainsTT is calculatedIt is unknown,
Again by r, the relationship between t obtains RIt is unknown, finally according to Y=RQTObtain YIt is unknown；The calculating process of step S10 is integrated, obtains one
A coefficient vector b or coefficient matrix B.
6. the inexpensive near-infrared spectrum method in identification medicinal tablet source according to claim 5, which is characterized in that will
The near infrared spectrum of the unknown medicinal tablet sample of acquisition projects in the coefficient vector of the near-infrared analysis model, identifies institute
The method for stating unknown medicinal tablet are as follows:
Unknown medicinal tablet sample spectra X is calculated according to y=Xb or Y=XBIt is unknown, wherein b is coefficient vector, B coefficient matrix.
7. the inexpensive near-infrared spectrum method in identification medicinal tablet source according to claim 5, it is characterised in that: adopt
Near-infrared analysis factor of a model number, PRESS are controlled with prediction residual square PRESS cooperation cross validation is defined as:
Wherein, n is training set sample number, f be because of subnumber,yijIt is the predicted value and reference value of sample label.
8. the inexpensive near-infrared spectrum method in identification medicinal tablet source according to claim 1, it is characterised in that: institute
Stating near-infrared analysis system includes the small-sized integrating sphere and Minitype infrared spectrum connected by optical fiber.
9. the inexpensive near-infrared spectrum method in identification medicinal tablet source according to claim 1, it is characterised in that:
The inner surface of the small-sized integrating sphere is applied whether there is or not the white coating of the uniform diffuse reflection of wavelength selectivity, in ball on any direction
Illumination be equal.
10. the inexpensive near-infrared spectrum method in identification medicinal tablet source according to claim 1, it is characterised in that:
Spectra collection is set as 20ms by integrating sphere, InGaAs detector array, the time of integration, and spectra collection wavelength region is 900-
1700nm, the average time of spectrum are 2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811524070.1A CN109632696A (en) | 2018-12-13 | 2018-12-13 | A kind of inexpensive near-infrared spectrum method identifying medicinal tablet source |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811524070.1A CN109632696A (en) | 2018-12-13 | 2018-12-13 | A kind of inexpensive near-infrared spectrum method identifying medicinal tablet source |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109632696A true CN109632696A (en) | 2019-04-16 |
Family
ID=66073527
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811524070.1A Pending CN109632696A (en) | 2018-12-13 | 2018-12-13 | A kind of inexpensive near-infrared spectrum method identifying medicinal tablet source |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109632696A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110987857A (en) * | 2019-12-23 | 2020-04-10 | 辽宁上药好护士药业（集团）有限公司 | Near-infrared detection and identification method for attributes of wild ginseng under forest |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103760110A (en) * | 2013-05-23 | 2014-04-30 | 山东商业职业技术学院 | Method for rapidly identifying meat with different animal sources |
| CN104020128A (en) * | 2014-02-21 | 2014-09-03 | 中国农业科学院蜜蜂研究所 | Method for rapidly identifying propolis source |
| CN107255625A (en) * | 2017-06-19 | 2017-10-17 | 山东大学 | One kind is based on Embedded tea-leaf producing area identification device and its application |
| CN107607485A (en) * | 2017-08-30 | 2018-01-19 | 浙江医药高等专科学校 | A kind of method for differentiating the radix tetrastigme place of production |
| CN107907500A (en) * | 2017-10-19 | 2018-04-13 | 宁夏大学 | A kind of detection method in the mutton place of production |
| CN108444944A (en) * | 2018-02-24 | 2018-08-24 | 盐城工学院 | A kind of radix polygoni multiflori powder place of production discrimination method for the spectrometry that diffused based on near-infrared |
| CN108645809A (en) * | 2018-06-27 | 2018-10-12 | 广西民族大学 | A kind of method that near-infrared spectrum technique quickly identifies rosin original tree species |
-
2018
- 2018-12-13 CN CN201811524070.1A patent/CN109632696A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103760110A (en) * | 2013-05-23 | 2014-04-30 | 山东商业职业技术学院 | Method for rapidly identifying meat with different animal sources |
| CN104020128A (en) * | 2014-02-21 | 2014-09-03 | 中国农业科学院蜜蜂研究所 | Method for rapidly identifying propolis source |
| CN107255625A (en) * | 2017-06-19 | 2017-10-17 | 山东大学 | One kind is based on Embedded tea-leaf producing area identification device and its application |
| CN107607485A (en) * | 2017-08-30 | 2018-01-19 | 浙江医药高等专科学校 | A kind of method for differentiating the radix tetrastigme place of production |
| CN107907500A (en) * | 2017-10-19 | 2018-04-13 | 宁夏大学 | A kind of detection method in the mutton place of production |
| CN108444944A (en) * | 2018-02-24 | 2018-08-24 | 盐城工学院 | A kind of radix polygoni multiflori powder place of production discrimination method for the spectrometry that diffused based on near-infrared |
| CN108645809A (en) * | 2018-06-27 | 2018-10-12 | 广西民族大学 | A kind of method that near-infrared spectrum technique quickly identifies rosin original tree species |
Non-Patent Citations (1)
| Title |
|---|
| 周孟然: "《煤矿瓦斯的激光光谱检测技术研究》", 31 May 2012 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110987857A (en) * | 2019-12-23 | 2020-04-10 | 辽宁上药好护士药业（集团）有限公司 | Near-infrared detection and identification method for attributes of wild ginseng under forest |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Biancolillo et al. | Chemometric methods for spectroscopy-based pharmaceutical analysis | |
| Khoshayand et al. | Simultaneous spectrophotometric determination of paracetamol, ibuprofen and caffeine in pharmaceuticals by chemometric methods | |
| Pelletier | Quantitative analysis using Raman spectrometry | |
| Chen et al. | Classification of different animal fibers by near infrared spectroscopy and chemometric models | |
| US10151688B2 (en) | Methodology for the identification of materials through methods of comparison of the spectrum of a sample against a reference library of spectra of materials | |
| Sparén et al. | Matrix effects in quantitative assessment of pharmaceutical tablets using transmission Raman and near-infrared (NIR) spectroscopy | |
| Zhang et al. | Evaluating the effects of causes of death on postmortem interval estimation by ATR-FTIR spectroscopy | |
| Grabska et al. | Theoretical simulation of near-infrared spectrum of piperine: insight into band origins and the features of regression models | |
| Sabin et al. | Characterization of sildenafil citrate tablets of different sources by near infrared chemical imaging and chemometric tools | |
| Chapman et al. | Spectroscopic approaches for rapid beer and wine analysis | |
| Melucci et al. | Rapid in situ repeatable analysis of drugs in powder form using reflectance near‐infrared spectroscopy and multivariate calibration | |
| Haware et al. | Data fusion of Fourier transform infrared spectra and powder X-ray diffraction patterns for pharmaceutical mixtures | |
| Hoffmann et al. | Spectra transfer between a Fourier transform near-infrared laboratory and a miniaturized handheld near-infrared spectrometer | |
| Deconinck et al. | An infrared spectroscopic approach to characterise white powders, easily applicable in the context of drug checking, drug prevention and on‐site analysis | |
| Vu Dang et al. | Chemometrics-based spectroscopy for pharmaceutical and biomedical analysis | |
| Uddin et al. | Chemometrics assisted spectrophotometric method for simultaneous determination of paracetamol and caffeine in pharmaceutical formulations | |
| Awotunde et al. | Discrimination of substandard and falsified formulations from genuine pharmaceuticals using NIR spectra and machine learning | |
| Zhang et al. | Rapid identification of the storage age of dried tangerine peel using a hand-held near infrared spectrometer and machine learning | |
| Lan et al. | Rapid evaluation on pharmacodynamics of Curcumae Rhizoma based on Micro-NIR and benchtop-NIR | |
| John et al. | Overview of cocaine identification by vibrational spectroscopy and chemometrics | |
| CN109632696A (en) | A kind of inexpensive near-infrared spectrum method identifying medicinal tablet source | |
| USMAN et al. | Applications of miniaturized and portable near infrared (NIR), Fourier transform infrared (FT-IR) and Raman spectrometers for the inspection and control of pharmaceutical products | |
| Kreft et al. | Qualitative determination of polyvinylpyrrolidone type by near-infrared spectrometry | |
| Muthudoss et al. | Machine learning-enabled NIR spectroscopy in assessing powder blend uniformity: clear-up disparities and biases induced by physical artefacts | |
| Monakhova et al. | Retrospective multivariate analysis of pharmaceutical preparations using 1H nuclear magnetic resonance (NMR) spectroscopy: example of 990 heparin samples |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190416 |
|
| RJ01 | Rejection of invention patent application after publication |