CN109374569A - A kind of detection method of the testing melamine content in milk powder based on tera-hertz spectra - Google Patents
A kind of detection method of the testing melamine content in milk powder based on tera-hertz spectra Download PDFInfo
- Publication number
- CN109374569A CN109374569A CN201811127592.8A CN201811127592A CN109374569A CN 109374569 A CN109374569 A CN 109374569A CN 201811127592 A CN201811127592 A CN 201811127592A CN 109374569 A CN109374569 A CN 109374569A
- Authority
- CN
- China
- Prior art keywords
- milk powder
- melamine
- model
- sample
- terahertz light
- 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
- JDSHMPZPIAZGSV-UHFFFAOYSA-N melamine Chemical compound NC1=NC(N)=NC(N)=N1 JDSHMPZPIAZGSV-UHFFFAOYSA-N 0.000 title claims abstract description 97
- 238000001228 spectrum Methods 0.000 title claims abstract description 97
- 229920000877 Melamine resin Polymers 0.000 title claims abstract description 96
- 239000000843 powder Substances 0.000 title claims abstract description 93
- 235000013336 milk Nutrition 0.000 title claims abstract description 92
- 239000008267 milk Substances 0.000 title claims abstract description 92
- 210000004080 milk Anatomy 0.000 title claims abstract description 92
- 238000001514 detection method Methods 0.000 title claims abstract description 37
- 238000012360 testing method Methods 0.000 title claims abstract description 26
- 230000000694 effects Effects 0.000 claims abstract description 22
- 238000000034 method Methods 0.000 claims description 39
- 238000010521 absorption reaction Methods 0.000 claims description 37
- 238000012937 correction Methods 0.000 claims description 15
- 238000012216 screening Methods 0.000 claims description 12
- 230000008033 biological extinction Effects 0.000 claims description 11
- 238000010606 normalization Methods 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 8
- 238000004611 spectroscopical analysis Methods 0.000 claims description 7
- 238000001328 terahertz time-domain spectroscopy Methods 0.000 claims description 6
- XZMCDFZZKTWFGF-UHFFFAOYSA-N Cyanamide Chemical compound NC#N XZMCDFZZKTWFGF-UHFFFAOYSA-N 0.000 claims description 4
- 238000004422 calculation algorithm Methods 0.000 claims description 4
- 239000000203 mixture Substances 0.000 claims description 4
- 238000011156 evaluation Methods 0.000 claims description 3
- 238000005259 measurement Methods 0.000 claims description 3
- VYPSYNLAJGMNEJ-UHFFFAOYSA-N Silicium dioxide Chemical compound O=[Si]=O VYPSYNLAJGMNEJ-UHFFFAOYSA-N 0.000 claims description 2
- 238000001035 drying Methods 0.000 claims description 2
- 230000005484 gravity Effects 0.000 claims description 2
- 239000004570 mortar (masonry) Substances 0.000 claims description 2
- 238000005303 weighing Methods 0.000 claims description 2
- 230000003595 spectral effect Effects 0.000 abstract description 19
- 238000005516 engineering process Methods 0.000 abstract description 4
- 238000009659 non-destructive testing Methods 0.000 abstract description 3
- 238000007781 pre-processing Methods 0.000 abstract description 3
- 230000003287 optical effect Effects 0.000 description 11
- 238000012843 least square support vector machine Methods 0.000 description 10
- 238000010586 diagram Methods 0.000 description 8
- 238000002203 pretreatment Methods 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 4
- 235000013305 food Nutrition 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 3
- 239000012071 phase Substances 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 241001269238 Data Species 0.000 description 2
- 238000002835 absorbance Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000004166 bioassay Methods 0.000 description 2
- 230000005684 electric field Effects 0.000 description 2
- 239000002778 food additive Substances 0.000 description 2
- 235000013373 food additive Nutrition 0.000 description 2
- 238000005064 physico chemical analysis method Methods 0.000 description 2
- 235000008476 powdered milk Nutrition 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 238000005829 trimerization reaction Methods 0.000 description 2
- VYZAMTAEIAYCRO-UHFFFAOYSA-N Chromium Chemical compound [Cr] VYZAMTAEIAYCRO-UHFFFAOYSA-N 0.000 description 1
- 150000001412 amines Chemical class 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000000149 argon plasma sintering Methods 0.000 description 1
- 238000003556 assay Methods 0.000 description 1
- 238000004587 chromatography analysis Methods 0.000 description 1
- 239000006071 cream Substances 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000004817 gas chromatography Methods 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000003018 immunoassay Methods 0.000 description 1
- 239000004615 ingredient Substances 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
- 239000007791 liquid phase Substances 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 150000007974 melamines Chemical class 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000010238 partial least squares regression Methods 0.000 description 1
- 235000020185 raw untreated milk Nutrition 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000010187 selection method Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 230000009466 transformation 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
- G01N21/3581—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light using far infrared light; using Terahertz radiation
- G01N21/3586—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light using far infrared light; using Terahertz radiation by Terahertz time domain spectroscopy [THz-TDS]
Abstract
The invention discloses a kind of detection methods of testing melamine content in milk powder based on tera-hertz spectra, comprising the following steps: (1) configures the milk powder sample containing preset concentration melamine；(2) each sample is analyzed using terahertz light spectra system, obtains the terahertz light spectrum information of each sample；(3) terahertz light spectrum information is pre-processed；(4) model between melamine concentration and terahertz light spectrum information is established；(5) the terahertz light spectrum information of milk powder to be detected is obtained, and is input in the model, to obtain the concentration of melamine in milk powder to be detected.The present invention uses terahertz light spectral technology, can be realized non-destructive testing, and detection speed is fast, in addition, interference can be reduced by pre-processing to terahertz light spectrum information, pass through the model established between melamine concentration and terahertz light spectrum information, it can be realized high-precision detection, detection effect is good.
Description
Technical field
The present invention relates to technical field of chemical detection, more particularly to melamine in a kind of milk powder based on tera-hertz spectra
The detection method of amine content.
Background technique
Food safety and each of us life are closely bound up, according to food additives GB2760-2014 national standard
Regulation, it is incorrect human health to be caused to seriously endanger using food additives.In recent years, in the food frequently broken out
In quality security problem, protrusion is exactly quality of milk powder safety the most, and the melamine in milk powder directly affects food safety.
The Limited Doses of melamine are 1mg/kg in milk power for infant and young children, and the product higher than 1mg/kg must not be sold without exception
It sells.Liquid milk (including raw milk), milk powder, the Limited Doses of melamine are 2.5mg/kg in other prescription emulsifiable powders, are higher than
The product of 2.5mg/kg must not be sold without exception.The Limited Doses of melamine are 2.5mg/ in other food containing 15% or more cream
Kg, the product higher than 2.5mg/kg must not be sold without exception.
The method of detection testing melamine content in milk powder is more at present, commonly includes bioassay method and physico-chemical analysis
Method, wherein bioassay method includes immunoassay and hexavalent chrome bio-removal, and physico-chemical analysis method includes gas chromatography and liquid
Phase chromatography.These detection methods although have the advantages that it is certain, but mostly to damage detection, and detection process complexity, period
Long problem.
Summary of the invention
In view of the above situation, the purpose of the present invention is to provide melamines in a kind of milk powder based on tera-hertz spectra to contain
The detection method of amount damages detection, and the problem that detection process is complicated, the period is long to solve the prior art.
A kind of detection method of the testing melamine content in milk powder based on tera-hertz spectra, comprising the following steps:
(1) the milk powder sample containing preset concentration melamine is configured；
(2) each sample is analyzed using terahertz light spectra system, obtains the terahertz light spectrum information of each sample；
(3) terahertz light spectrum information is pre-processed；
(4) model between melamine concentration and terahertz light spectrum information is established；
(5) the terahertz light spectrum information of milk powder to be detected is obtained, and is input in the model, to obtain milk powder to be detected
The concentration of middle melamine.
Compared with prior art, detection method provided by the invention can at normal temperature using terahertz light spectral technology
Sample can be detected, detection speed is fast, nor needs to destroy sample and cause the consumption of sample, Neng Goushi
Existing non-destructive testing, in addition, interference can be reduced by pre-processing to terahertz light spectrum information, it is dense by establishing melamine
Model between degree and terahertz light spectrum information, can be realized high-precision detection, detection effect is good.
Wherein, in the step (1) in each sample the mass percent of melamine range are as follows: 0.5%~20%.
Wherein, 0.5% is divided between the concentration gradient of the melamine of each sample in the step (1).
Wherein, the step (2) specifically includes:
Each sample is analyzed using terahertz time-domain spectroscopy instrument, the scanning range of the terahertz time-domain spectroscopy instrument
For 1.0-5.0THz, resolution ratio 7.6THz, scanning times are set as 8192 times, then obtain each sample 1.0~
The terahertz light spectrum information of 3.0THz.
Wherein, the tera-hertz spectra information includes time-domain spectroscopy information, and time-domain spectroscopy information is become through Fourier
The frequency domain spectra information obtained after changing.
Wherein, the terahertz light spectrum information specifically includes absorption coefficient, extinction coefficient, refractive index.
Wherein, the step (3) includes:
Place is corrected to the terahertz light spectrum information of acquisition using smooth, multiplicative scatter correction, baseline correction, normalization
Reason；
Partial least square model is established respectively, and the partial least square model by establishing comments pretreated effect
Valence, to obtain best preprocess method.
Wherein, the step (3) specifically includes:
Sample is divided into modeling collection and forecast set, the modeling collects and the quantity of the forecast set is than being about 3:1, described to build
Mould collection is for establishing model, and the verifying collection is for the verifying to model；
Collected terahertz light spectrum information is corrected using smooth, multiplicative scatter correction, baseline correction, normalization
Processing；
Partial least square model is established respectively, and the partial least square model by establishing comments pretreated effect
Valence, to obtain best preprocess method, wherein referred to using prediction related coefficient and predicted root mean square error as the evaluation of model
Mark.
Wherein, the model established in the step (4) is partial least square model or least square method supporting vector machine model.
Wherein, before the step (4), the method also includes:
Wave band is carried out using successive projection algorithm or without information variable null method to pretreated terahertz light spectrum information
Screening.
Detailed description of the invention
Above-mentioned and/or additional aspect of the invention and advantage will become from the description of the embodiment in conjunction with the following figures
Obviously and it is readily appreciated that, in which:
Fig. 1 is the absorbance comparison diagram of the milk powder of the melamine containing various concentration；
Fig. 2 is the reflection coefficient comparison diagram of the milk powder of the melamine containing various concentration；
Fig. 3 is the absorption coefficient comparison diagram of the milk powder of the melamine containing various concentration；
Fig. 4 is the refraction coefficient comparison diagram of the milk powder of the melamine containing various concentration；
Fig. 5 is terahertz light spectral absorption coefficient modeling collection root-mean-square error and forecast set root-mean-square error with the principal component factor
The curve graph of number variation；
Fig. 6 is the regression coefficient curve graph of terahertz light spectral absorption coefficient；
Fig. 7 is the doping content of melamine predicted value and true value fitted figure established using Terahertz absorption coefficient；
Fig. 8 is absorption coefficient UVE method variable preferred result figure；
Fig. 9 is that SPA models Input variable selection decision figure；
Figure 10 is the predicted value and true value scatter plot using Terahertz absorption coefficient to testing melamine content in milk powder.
Specific embodiment
To keep objects, features and advantages of the present invention more obvious and easy to understand, with reference to the accompanying drawing to tool of the invention
Body embodiment is described in detail.Several embodiments of the invention are given in attached drawing.But the present invention can be with many not
With form realize, however it is not limited to embodiment described herein.On the contrary, purpose of providing these embodiments is makes to this
The disclosure of invention is more thorough and comprehensive.
Unless otherwise defined, all technical and scientific terms used herein and belong to technical field of the invention
The normally understood meaning of technical staff is identical.Term as used herein in the specification of the present invention is intended merely to description tool
The purpose of the embodiment of body, it is not intended that in the limitation present invention.Term " and or " used herein includes one or more phases
Any and all combinations of the listed item of pass.
A kind of detection method of the testing melamine content in milk powder based on tera-hertz spectra, comprising the following steps:
(1) the milk powder sample containing preset concentration melamine is configured；
(2) each sample is analyzed using terahertz light spectra system, obtains the terahertz light spectrum information of each sample；
(3) terahertz light spectrum information is pre-processed；
(4) model between melamine concentration and terahertz light spectrum information is established；
(5) the terahertz light spectrum information of milk powder to be detected is obtained, and is input in the model, to obtain milk powder to be detected
The concentration of middle melamine.
Wherein, step (1) specifically includes:
The melamine powder of preset concentration is poured into agate mortar and is ground into 2 μm of fine powders, is subsequently placed in baking oven and toasts
4-5h, the sample of drying are put into drier for use；
The milk powder for weighing corresponding specific gravity, puts it into centrifuge tube, labelled and cover bottle cap；
Centrifuge tube equipped with melamine and milk powder is placed in vortex mixer, mixed 3min is shaken, until mix well for
Only；
The mixture of melamine and milk powder after mixing is put into mold using tablet press machine, it is made to be pressed into diameter
1cm, the tabletting with a thickness of 1mm；
Using digital display type vernier caliper, the thickness of the sample pressed is recorded the parameter by measurement, by configured sample
Originally it is placed on spare in print frame.
Wherein, milk powder can be bought from supermarket, and the purity assay of the melamine of sample should be greater than or be equal to 99.7%.
In the present embodiment, sample is prepared, concentration range is in 0.5%- when preparing according to 0.5% gradient
Between 20%.First according to corresponding gradient, concentration proportioning table is formulated, wherein by the same way, prepare 41 various concentrations altogether
Milk powder and melamine concentration mix sample.Each concentration prepares 4 samples, and milk powder is obtained and melamine mixing sample is each
164, the milk powder and melamine concentration table of prepared 41 gradients are as shown in table 1.
The mass percent (C%) of various concentration melamine sample is adulterated in 1 milk powder of table
Wherein, the true value distribution statistics result such as table of the milk powder sample collection of different content of melamine and forecast set
Shown in 2.
Table 2 adulterates the milk powder sample collection and forecast set true value distribution statistics table of various concentration melamine
Wherein, the step (2) specifically includes:
Each sample is analyzed using terahertz time-domain spectroscopy instrument, the scanning range of the terahertz time-domain spectroscopy instrument
For 1.0-5.0THz, resolution ratio 7.6THz, scanning times are set as 8192 times, then obtain each sample 1.0~
The terahertz light spectrum information of 3.0THz.It wherein, is the influences of the factors to experimental result such as reduction sample configuration, tabletting, each milk
Powder sample acquires 5 tera-hertz spectras, and finds out average tera-hertz spectra for establishing quantifying for testing melamine content in milk powder
Detection model.
In the present embodiment, the tera-hertz spectra information includes time-domain spectroscopy information, and by time-domain spectroscopy information through Fu
In the frequency domain spectra information that obtains after leaf transformation.
Specifically, the terahertz light spectrum information specifically includes absorption coefficient, extinction coefficient, refractive index.
That detect in THz wave is the function E (t) that changes over time of electric field of terahertz pulse, to Terahertz when
Domain waveform carries out Fast Fourier Transform (FFT) and obtains frequency-domain waveform:
In formula, ω is frequency, and t is the time, by formula (1-1) it is found that the electric field strength of Terahertz includes simultaneously in frequency domain
The information of amplitude and phase, can very easily extract material in the optical parameter of this frequency range.
The refractive index calculation formula of sample are as follows:
The extinction coefficient calculation formula of sample are as follows:
The extinction coefficient calculation formula of sample are as follows:
In addition, its calculation formula can be derived by Lambert-Beer's law in an experiment, it is quickly obtained the absorbing state of sample
Are as follows:
α=[- Ln (IsampleIreference)]d
Wherein, IsampleRepresent the terahertz pulse light intensity for having sample message, IreferenceRepresent the terahertz of reference signal
Hereby light intensity, the thickness of d representative sample.
In step (2), the abundant information of tera-hertz spectra, since it is with transmitance, absorption coefficient, extinction coefficient, folding
The terahertz optics parameter such as rate, dielectric constant, phase angle is penetrated, the internal information of substance can be reflected with various dimensions.To milk powder and three
Poly cyanamid sterling spectrum carries out detailed analysis.Such as the different optics that Fig. 1 and Fig. 2 is milk powder and melamine sample tera-hertz spectra
Parameter comparison, for convenient for later data, intercepting frequency range herein is 1.0-3.0THz frequency range.
Wherein, Fig. 1 is the absorbance comparison diagram of the milk powder of the melamine containing various concentration, and sterling milk powder has in the position 1.4T
One more apparent peak value, the peak value be due to milk powder inside the influence of component generate, with the increase of Terahertz frequency,
Doping powdered milk sample absorption coefficient curve slowly rises.The melamine of sterling has apparent peak between 2.029T and 2.281T
Value；Fig. 2 is the reflection coefficient comparison diagram of the milk powder of the melamine containing various concentration, wherein the figure of pure milk powder close to straight line,
Pure melamine has turning point in the position of 2.251T.If table 3 is that milk powder and melamine sterling spectral signature compare.
Table 3 is that milk powder and melamine sterling spectral signature compare
Please refer to Fig. 3 and Fig. 4, respectively the absorption coefficient comparison diagram of the milk powder of the melamine containing various concentration and containing difference
The refraction coefficient comparison diagram of the milk powder of concentration melamine, by upper figure it is found that respectively for the spectrum under various concentration in milk powder
The peak position of data, various concentration parameter is coincide, and is illustrated that experimental data is reliable, is also demonstrated the reproducible of experiment.Wherein, from figure
3 waveform can be seen that the powdered milk sample Terahertz absorption coefficient original spectrum useful information of doping various concentration melamine
It is concentrated mainly in 1.0-3.0THz band limits, since there are noise informations at spectrum both ends, therefore chooses 1.0- when modeling
3.0THz wave band.There is stronger absorption peak in the position of 2.029T, 2.281T, 2.647T, and with melamine concentration
Increase, intensity gradually increases.Due to the influence of the ingredient in milk powder, there is an absorption peak of milk powder in the position 1.404T.It is overall
The original tera-hertz spectra waveform of various concentration melamine is consistent in milk powder, and only the intensity in certain wave band has
Not.Due to being mixed with high concentration melamine in milk powder, so that sample is more strong to the absorption of Terahertz wave spectrum.
Can be seen that from the waveform of Fig. 4 has a fluctuation in the position of 2.251T, and with the increasing of melamine concentration
Greatly, refractive index integral strength gradually increases.The original tera-hertz spectra waveform of various concentration melamine is consistent in overall milk powder
, only the intensity in certain wave band is had any different.Due to being mixed with melamine in milk powder the refractive index of sample is occurred
Variation, and melamine concentration is higher, and the parameter of refractive index is bigger
Wherein, the step (3) includes:
Place is corrected to the terahertz light spectrum information of acquisition using smooth, multiplicative scatter correction, baseline correction, normalization
Reason；
Partial least square model is established respectively, and the partial least square model by establishing comments pretreated effect
Valence, to obtain best preprocess method.
When it is implemented, the step (3) specifically includes:
Sample is divided into modeling collection and forecast set, the modeling collects and the quantity of the forecast set is than being about 3:1, described to build
Mould collection is for establishing model, and the verifying collection is for the verifying to model；
Collected terahertz light spectrum information is corrected using smooth, multiplicative scatter correction, baseline correction, normalization
Processing；
Partial least square model is established respectively, and the partial least square model by establishing comments pretreated effect
Valence, to obtain best preprocess method, wherein referred to using prediction related coefficient and predicted root mean square error as the evaluation of model
Mark.
In order to reduce laboratory apparatus because the reasons such as vibration, noise and temperature cause original spectrum to generate drift, sample not
Uniformly, light scattering etc. influences, appropriate pretreatment is carried out to the terahertz light spectrum information of acquisition it is necessary to, suitable pretreatment with
Obtain preferably modeling effect.Following bearing calibration, such as smooth (Smoothing), multiplicative scatter correction are mainly used herein
(MSC), baseline correction (Baseline), normalization (Normalization) etc., carry out school to the terahertz light spectrum information of acquisition
Partial least square model is established after positive processing respectively, and the partial least square model (PLS) by establishing is to pretreated effect
Fruit is evaluated.
PLS model is established after each optical parameter different pretreatments of the tera-hertz spectra of testing melamine content in milk powder.It participates in
The terahertz light modal data of modeling has 160, and the major optical parameter of Terahertz includes absorption coefficient, extinction coefficient and refractive index
Deng and 160 melamine concentration true value as input.Using chemometrics method to tera-hertz spectra data information into
5 kinds of distinct methods of row are pre-processed, and are established quantitative model in conjunction with offset minimum binary (PLS) and evaluated.Wherein, above-mentioned
160 terahertz light modal datas are divided into 120 calibration sets, 40 forecast sets.And by comparing the modeling collection of PLS model and prediction
The coefficient R of collectionC、RP, RMSEC, RMSEP evaluate 5 kinds of pretreated effects.Table 4 be testing melamine content in milk powder too
PLS model is established after hertz each optical parameter different pretreatments of spectrum, is analyzed using the absorption coefficient of tera-hertz spectra, is passed through
It crosses and normalizes that pretreated effect is preferable, the PLS model modeling collection r of foundationcIt is 0.991, modeling collection RMSEC is 0.008, prediction
Collect rpIt is 0.987, prediction RMSEP is 0.010；
PLS model is established after each optical parameter different pretreatments of the tera-hertz spectra of 4 testing melamine content in milk powder of table
PLS model is established after each optical parameter different pretreatments of the tera-hertz spectra of testing melamine content in milk powder.It participates in
The terahertz light modal data of modeling has 160, and the major optical parameter of Terahertz includes absorption coefficient, extinction coefficient and refractive index
Deng and 160 melamine concentration true value as input.Using chemometrics method to tera-hertz spectra data information into
5 kinds of distinct methods of row are pre-processed, and are established quantitative model in conjunction with offset minimum binary (PLS) and evaluated.Wherein, above-mentioned
160 terahertz light modal datas are divided into 120 calibration sets, 40 forecast sets.And by comparing the modeling collection of PLS model and prediction
Coefficient R C, RP, RMSEC, RMSEP of collection evaluate 5 kinds of pretreated effects.Table 4 is testing melamine content in milk powder
PLS model is established after each optical parameter different pretreatments of tera-hertz spectra, is analyzed using the absorption coefficient of tera-hertz spectra,
Preferable, the PLS model modeling collection r of foundation by the pretreated effect of normalizationcIt is 0.991, modeling collection RMSEC is 0.008, in advance
Survey collection rpIt is 0.987, prediction RMSEP is 0.010.
The tera-hertz spectra of modeling collection and forecast set establishes mould for subsequent after above-mentioned best pretreatment mode processing
Type.We are by taking absorption coefficient as an example, to introduce modeling process.Principal component is because subnumber (PC) is excessively high or too low can all influence number
The precision of model is learned, if the principal component of selection is lower because of subnumber (PC), it is likely that lose effective spectral information；If the master of selection
Components Factor number (PC) is higher, then model is easy to appear overfitting phenomenon.If Fig. 5 is terahertz light spectral absorption coefficient modeling collection
The curve graph that root-mean-square error and forecast set root-mean-square error change with principal component by subnumber.In the milk powder trimerization that this experiment is established
In the PLS model of the terahertz light spectral absorption coefficient of cyanamide content, the best principal component selected is because subnumber is 4.
Such as the regression coefficient curve graph that Fig. 6 is terahertz light spectral absorption coefficient, calculated back by analysis model recurrence
Return coefficient, regression coefficient and spectral variables are added into intercept b=0.0768 by weighted sum, trimerization in milk powder can be obtained
The quantitative detection model of cyanamide content, if Fig. 7 is the doping content of melamine predicted value established using Terahertz absorption coefficient
With true value fitted figure.
In addition, in the present embodiment, before the step (4), the method also includes:
To pretreated terahertz light spectrum information using successive projection algorithm (SPA) or without information variable null method
(UVE) wave band screening is carried out.
Since tera-hertz spectra data volume is larger, using full wave terahertz light modal data as input variable, data
Time-consuming for more, data processing.Wavelength screening technique appropriate is selected, the modeling time can not only be shortened, moreover it is possible to guarantee modeling
Precision.In addition, removing superfluous in order to avoid the noise of spectrometer and some wavelength change sensitive influence to outside environmental elements
Remaining information can greatly improve the processing speed of data.
UVE parameter selection: carrying out wave band screening to spectrum is a kind of very effective spectral manipulation method, can be significantly excellent
Change model, wave band screening is carried out by no information variable null method, the principle of this method is the wave established based on PLS regression coefficient
Section screening technique, using regression coefficient as important measurement index, spectrum square is added in the stochastic variable matrix of certain amount by this method
In battle array, then PLS model established by validation-cross, by calculate the average value of PLS and regression coefficient and standard deviation ratio come
Effective spectral information is chosen, noise signal and concentration information are integrated in one by this method, convenient and intuitive.In addition, UVE wave band screens
Data redundancy can be removed very well, improve arithmetic speed.The screening of UVE wave band is carried out in tera-hertz spectra is composed entirely, it is vertical vertical
Solid line is the segmentation boundary of wavelength variable and random noise variable, and the left side is wavelength variable, and the right is noise variance.
SPA parameter selection: successive projection algorithm is according to the rectangular projection information content of Terahertz wavelength variable come from Terahertz
In spectrum parameter variable, select the absorption coefficient, extinction coefficient, refractive index data of testing melamine content in milk powder as defeated respectively
Enter.It is respectively 35 and 10 that SPA, which is arranged, to choose variable number, and SPA wave band the selection result is illustrated in fig. 9 shown below.Adulterate the milk of melamine
UVE the and SPA feature extraction of powder sample terahertz light spectral absorption coefficient parameter.
Wherein, Fig. 8 is absorption coefficient UVE method variable preferred result figure, carries out variable optimization using UVE method, shares
103 wavelength variables are selected.The wavelength points information contained for using UVE wave band screening technique to filter out is established more as input variable
Add simplified mathematical model.Fig. 9 is that SPA models Input variable selection decision figure.Wherein, melamine sample is adulterated in 5 milk powder of table
The absorption coefficient optical parameter of product Terahertz carries out SPA wave band and screens effective information variable.Terahertz absorption coefficient is by operation
Afterwards, it after 13 wavelength variables are chosen as input vector, is modeled through PLS, root-mean-square error 1.32%.
Melamine sample tera-hertz spectra different parameters SPA is adulterated in 5 milk powder of table screens useful variable
In the present embodiment, the model established in the step (4) is partial least square model (PLS) or least square is supported
Vector machine model (LS-SVM).
Offset minimum binary is statistical method, and the information that this method can make full use of small sample to include will make full use of
The linear relationship between melamine milk powder sample terahertz light modal data and content of melamine is adulterated to establish model.Table 6
Modeling effect for different wave length screening technique combination PLS model compares.
The PLS modelling effect that the different Variable Selection methods of table 6 establish the screening of tera-hertz spectra difference optical parameter compares
As can be drawn from Table 6, it is modeled using three kinds of optical parameters of Terahertz of the milk powder sample of doping melamine, every kind
Parameter respectively corresponds full spectrum PLS, SPA-PLS, UVE-PLS model, wherein models for Terahertz absorption coefficient, full spectrum modeling
Effect is best, forecast set R2It is respectively 0.987,0.001 with RMSEP, in addition, SPA picks out 13 wavelength in SPA-PLS
Number, forecast set R2Still reach 0.966,0.013 with RMSEP.
Partial Least Square Support Vector Machine (LS-SVM) key index has input vector, kernel function and its parameter.LS-SVM
Two kinds of typical kernel functions be respectively Radial basis kernel function and linear kernel function.As shown in Table 7, Radial basis kernel function model result
It is whole to be better than linear kernel function.It is stronger that reason is likely to Radial basis kernel function generalization ability, can approach any non-linear letter
Number can handle doping melamine testing melamine content in milk powder true value and doping melamine milk powder sample Terahertz well
Non-linear relation between spectroscopic data.Wherein, melamine sample terahertz light spectral absorption coefficient parameter combines not in 7 milk powder of table
With kernel function LS-SVM modeling result.
7 milk powder of table adulterates melamine sample terahertz light spectral absorption coefficient modeling result
From upper table 7 it is found that the terahertz light spectral absorption coefficient of milk powder doping melamine sample combines different kernel function LS-
In SVM modeling, the results showed that when using Lin kernel function, in SPA-LSSVM, model parameter γ=6.226, forecast set R2
It is respectively 0.985,0.010 with RMSEP, when using RBF kernel function, there are two important parameters: γ and σ2, the two unknown parameters
Model generalization ability and modeling effect are influenced in a sense.Two parameters of model finally are determined, at this time SPA-LSSVM
The optimal combination of model parameter are as follows: γ=1.724e+03, σ2=3.458, forecast set R2Be respectively 0.996 with RMSEP,
0.005。
If Figure 10 is the predicted value and true value scatter plot using Terahertz absorption coefficient to testing melamine content in milk powder,
The LS-SVM model prediction collection Rp of middle foundation is 0.996, and forecast set RMSEP is 0.005.
Method provided in this embodiment uses terahertz light spectral technology, and selecting three kinds of parameters of Terahertz, (absorption coefficient disappears
Backscatter extinction logarithmic ratio, refractive index), the effect of modeling is all generally speaking preferable, but is directed to different parameters, optimal pretreatment side
Method, wave band screening and modeling method be not identical.For absorption coefficient, PLS and LS-SVM model is compared, it can be found that:
The SPA-LS-SVM forecast of regression model result established after normalization is best, predicted root mean square error 0.50%；For
For extinction coefficient, PLS and LS-SVM regression model is compared, it can be found that: the LS-SVM established after normalization processing is returned
Return the prediction effect of model best, predicted root mean square error 0.40%；For refractive index, PLS and LS-SVM mould is compared
Type, it can be found that: SPA-LS-SVM the and UVE-LS-SVM forecast of regression model effect difference established after baseline correction is little, leads to
Crossing UVE-LS-SVM regression model can make input data less, predicted root mean square error 0.55%.
In conclusion the prior art is compared, detection method provided by the invention, using terahertz light spectral technology, Neng Gou
Sample can be detected under room temperature, detection speed is fast, nor needs to destroy sample and sample is caused to disappear
Consumption, can be realized non-destructive testing, in addition, can reduce interference by pre-processing to terahertz light spectrum information, passes through foundation
Model between melamine concentration and terahertz light spectrum information, can be realized high-precision detection, and detection effect is good.
The embodiments described above only express several embodiments of the present invention, and the description thereof is more specific and detailed, but simultaneously
Limitations on the scope of the patent of the present invention therefore cannot be interpreted as.It should be pointed out that for those of ordinary skill in the art
For, without departing from the inventive concept of the premise, various modifications and improvements can be made, these belong to guarantor of the invention
Protect range.Therefore, the scope of protection of the patent of the invention shall be subject to the appended claims.
Claims (10)
1. a kind of detection method of the testing melamine content in milk powder based on tera-hertz spectra, which is characterized in that including following step
It is rapid:
(1) the milk powder sample containing preset concentration melamine is configured；
(2) each sample is analyzed using terahertz light spectra system, obtains the terahertz light spectrum information of each sample；
(3) terahertz light spectrum information is pre-processed；
(4) model between melamine concentration and terahertz light spectrum information is established；
(5) the terahertz light spectrum information of milk powder to be detected is obtained, and is input in the model, to obtain three in milk powder to be detected
The concentration of poly cyanamid.
2. the detection method of the testing melamine content in milk powder according to claim 1 based on tera-hertz spectra, feature
It is, the step (1) specifically includes:
The melamine powder of preset concentration is poured into agate mortar and is ground into 2 μm of fine powders, is subsequently placed in baking oven and toasts 4-5h,
The sample of drying is put into drier for use；
The milk powder for weighing corresponding specific gravity, puts it into centrifuge tube, labelled and cover bottle cap；
Centrifuge tube equipped with melamine and milk powder is placed in vortex mixer, mixed 3min is shaken, until mixing well；
The mixture of melamine and milk powder after mixing is put into mold using tablet press machine, it is made to be pressed into diameter 1cm, thickness
Degree is the tabletting of 1mm；
Using digital display type vernier caliper, the thickness of the sample pressed is recorded the parameter, configured sample is put by measurement
It sets spare in print frame.
3. the detection method of the testing melamine content in milk powder according to claim 2 based on tera-hertz spectra, feature
It is, is divided into 0.5% between the concentration gradient of the melamine of each sample in the step (1).
4. the detection method of the testing melamine content in milk powder according to claim 1 based on tera-hertz spectra, feature
It is, the step (2) specifically includes:
Each sample is analyzed using terahertz time-domain spectroscopy instrument, the scanning range of the terahertz time-domain spectroscopy instrument is
1.0-5.0THz, resolution ratio 7.6THz, scanning times are set as 8192 times, then obtain each sample in 1.0~3.0THz
Terahertz light spectrum information.
5. the detection method of the testing melamine content in milk powder according to claim 1 based on tera-hertz spectra, feature
It is, the tera-hertz spectra information includes time-domain spectroscopy information, and obtains after time-domain spectroscopy information is fourier transformed
Frequency domain spectra information.
6. the detection method of the testing melamine content in milk powder according to claim 5 based on tera-hertz spectra, feature
It is, the terahertz light spectrum information specifically includes absorption coefficient, extinction coefficient, refractive index.
7. the detection method of the testing melamine content in milk powder according to claim 1 based on tera-hertz spectra, feature
It is, the step (3) includes:
Processing is corrected to the terahertz light spectrum information of acquisition using smooth, multiplicative scatter correction, baseline correction, normalization；
Partial least square model is established respectively, and the partial least square model by establishing evaluates pretreated effect,
To obtain best preprocess method.
8. the detection method of the testing melamine content in milk powder according to claim 7 based on tera-hertz spectra, feature
It is, the step (3) specifically includes:
Sample is divided into modeling collection and forecast set, the modeling collects and the quantity of the forecast set is than being about 3:1, the modeling collection
For establishing model, the verifying collection is for the verifying to model；
Place is corrected to collected terahertz light spectrum information using smooth, multiplicative scatter correction, baseline correction, normalization
Reason；
Partial least square model is established respectively, and the partial least square model by establishing evaluates pretreated effect,
To obtain best preprocess method, wherein the evaluation index using prediction related coefficient and predicted root mean square error as model.
9. the detection method of the testing melamine content in milk powder according to claim 1 based on tera-hertz spectra, feature
It is, the model established in the step (4) is partial least square model or least square method supporting vector machine model.
10. the detection method of the testing melamine content in milk powder according to claim 1 based on tera-hertz spectra, feature
It is, before the step (4), the method also includes:
Wave band screening is carried out using successive projection algorithm or without information variable null method to pretreated terahertz light spectrum information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811127592.8A CN109374569A (en) | 2018-09-27 | 2018-09-27 | A kind of detection method of the testing melamine content in milk powder based on tera-hertz spectra |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811127592.8A CN109374569A (en) | 2018-09-27 | 2018-09-27 | A kind of detection method of the testing melamine content in milk powder based on tera-hertz spectra |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109374569A true CN109374569A (en) | 2019-02-22 |
Family
ID=65402165
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811127592.8A Pending CN109374569A (en) | 2018-09-27 | 2018-09-27 | A kind of detection method of the testing melamine content in milk powder based on tera-hertz spectra |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109374569A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109883981A (en) * | 2019-02-26 | 2019-06-14 | 浙江大学 | A kind of detection method of the soil crude oil pollution degree based on terahertz light spectral technology |
| CN110867221A (en) * | 2019-11-08 | 2020-03-06 | 北京农业信息技术研究中心 | Method and device for determining soil lead content prediction model based on terahertz spectrum |
| CN110987853A (en) * | 2019-11-08 | 2020-04-10 | 北京农业信息技术研究中心 | Method and device for predicting soil lead pollution degree based on terahertz spectrum |
| CN111781159A (en) * | 2020-07-30 | 2020-10-16 | 浙江大学 | In-situ detection method and system for plant cadmium chelate |
| CN111965134A (en) * | 2020-08-13 | 2020-11-20 | 桂林电子科技大学 | Terahertz spectrum quantitative analysis method for rubber vulcanization accelerator mixture |
| CN113049525A (en) * | 2021-03-11 | 2021-06-29 | 华东交通大学 | Terahertz metamaterial absorber, preparation method and trace benzoic acid content detection method based on terahertz metamaterial absorber |
| CN113655023A (en) * | 2021-07-15 | 2021-11-16 | 上海理工大学 | Method for rapidly detecting low-concentration atrazine |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101487796A (en) * | 2009-02-24 | 2009-07-22 | 中国检验检疫科学研究院 | Method for measuring melamine content in solid example |
| CN102331409A (en) * | 2011-08-11 | 2012-01-25 | 上海华碧检测技术有限公司 | Method for quickly determining content of melamine in milk powder |
| CN106596464A (en) * | 2016-12-23 | 2017-04-26 | 天津农学院 | Near-infrared autocorrelation spectrum detection method for melamine doped in milk powder |
| CN106706555A (en) * | 2016-11-21 | 2017-05-24 | 无锡迅杰光远科技有限公司 | Milk powder determination method and system based on near infrared spectroscopy technology |
| CN108267421A (en) * | 2017-12-30 | 2018-07-10 | 深圳市太赫兹科技创新研究院有限公司 | The detection method and detecting system of melamine in a kind of milk powder |
-
2018
- 2018-09-27 CN CN201811127592.8A patent/CN109374569A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101487796A (en) * | 2009-02-24 | 2009-07-22 | 中国检验检疫科学研究院 | Method for measuring melamine content in solid example |
| CN102331409A (en) * | 2011-08-11 | 2012-01-25 | 上海华碧检测技术有限公司 | Method for quickly determining content of melamine in milk powder |
| CN106706555A (en) * | 2016-11-21 | 2017-05-24 | 无锡迅杰光远科技有限公司 | Milk powder determination method and system based on near infrared spectroscopy technology |
| CN106596464A (en) * | 2016-12-23 | 2017-04-26 | 天津农学院 | Near-infrared autocorrelation spectrum detection method for melamine doped in milk powder |
| CN108267421A (en) * | 2017-12-30 | 2018-07-10 | 深圳市太赫兹科技创新研究院有限公司 | The detection method and detecting system of melamine in a kind of milk powder |
Non-Patent Citations (2)
| Title |
|---|
| 李建蕊等: "太赫兹时域谱技术快速定性检测奶粉中的三聚氰胺", 《中国计量学院学报》 * |
| 胡军: "基于太赫兹时域光谱技术的奶粉添加剂检测研究", 《道客巴巴》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109883981A (en) * | 2019-02-26 | 2019-06-14 | 浙江大学 | A kind of detection method of the soil crude oil pollution degree based on terahertz light spectral technology |
| CN110867221A (en) * | 2019-11-08 | 2020-03-06 | 北京农业信息技术研究中心 | Method and device for determining soil lead content prediction model based on terahertz spectrum |
| CN110987853A (en) * | 2019-11-08 | 2020-04-10 | 北京农业信息技术研究中心 | Method and device for predicting soil lead pollution degree based on terahertz spectrum |
| CN110987853B (en) * | 2019-11-08 | 2022-02-11 | 北京农业信息技术研究中心 | Method and device for predicting soil lead pollution degree based on terahertz spectrum |
| CN111781159A (en) * | 2020-07-30 | 2020-10-16 | 浙江大学 | In-situ detection method and system for plant cadmium chelate |
| CN111965134A (en) * | 2020-08-13 | 2020-11-20 | 桂林电子科技大学 | Terahertz spectrum quantitative analysis method for rubber vulcanization accelerator mixture |
| CN113049525A (en) * | 2021-03-11 | 2021-06-29 | 华东交通大学 | Terahertz metamaterial absorber, preparation method and trace benzoic acid content detection method based on terahertz metamaterial absorber |
| CN113049525B (en) * | 2021-03-11 | 2022-05-17 | 华东交通大学 | Terahertz metamaterial absorber and trace benzoic acid detection method based on metamaterial |
| CN113655023A (en) * | 2021-07-15 | 2021-11-16 | 上海理工大学 | Method for rapidly detecting low-concentration atrazine |
| CN113655023B (en) * | 2021-07-15 | 2023-09-05 | 上海理工大学 | Method for detecting low-concentration atrazine rapidly |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109374569A (en) | A kind of detection method of the testing melamine content in milk powder based on tera-hertz spectra | |
| CN104062257B (en) | A kind of based on the method for general flavone content near infrared ray solution | |
| Ren et al. | Multi-variable selection strategy based on near-infrared spectra for the rapid description of dianhong black tea quality | |
| Huang et al. | Improved generalization of spectral models associated with Vis-NIR spectroscopy for determining the moisture content of different tea leaves | |
| CN109799207A (en) | The quantitative detecting method of talcum powder is mixed in the root of Dahurain angelica based on near-infrared spectrum analysis | |
| Hu et al. | Optimization of quantitative detection model for benzoic acid in wheat flour based on CARS variable selection and THz spectroscopy | |
| Lan et al. | A mid-level data fusion approach for evaluating the internal and external changes determined by FT-NIR, electronic nose and colorimeter in Curcumae Rhizoma processing | |
| CN104359853B (en) | Method and application using near infrared spectroscopy quick detection Uncaria | |
| Zhu et al. | Review of portable near infrared spectrometers: Current status and new techniques | |
| Ning et al. | Quantitative detection of zearalenone in wheat grains based on near-infrared spectroscopy | |
| CN103411895B (en) | Pseudo-near infrared spectrum identification method mixed by pearl powder | |
| CN105784628A (en) | Method for detecting chemical composition of soil organic matter with mid-infrared spectra | |
| CN104849233A (en) | Cereal freshness detecting method and device | |
| CN108760677A (en) | A kind of rhizoma pinellinae praeparata based on near-infrared spectrum technique mixes pseudo- discrimination method | |
| CN103091274A (en) | Method for determining content of water in Salvianolic acid for injection through near-infrared diffuse reflection spectrometry | |
| CN105758819A (en) | Method for detecting organic components of soil by utilizing near infrared spectrum | |
| Tugnolo et al. | Characterization of green, roasted beans, and ground coffee using near infrared spectroscopy: A comparison of two devices | |
| CN100557413C (en) | Ammonium perchlorate particle size's rapid analysis in the solid propellant | |
| Bin et al. | Detection of adulteration of kudzu powder by terahertz time-domain spectroscopy | |
| Hu et al. | Establishment and optimization of temperature compensation model for benzoic acid detection based on terahertz metamaterial | |
| Qin et al. | Prediction of Peking duck intramuscle fat content by near-infrared spectroscopy | |
| CN1982873B (en) | Near-infrared diffuse reflection spectral method for fastly inspecting drop appearance quality | |
| Hu et al. | Predicting the moisture content of Daqu with hyperspectral imaging | |
| Suhandy et al. | The Measurement of Luwak Content in Coffee Blends Using UV-Visible Spectroscopy Combined with Support Vector Machine Regression (SVMR) | |
| CN105784629A (en) | Method for quick detection of stable carbon isotope ratio of soil by means of mid-infrared spectroscopy |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190222 |
|
| RJ01 | Rejection of invention patent application after publication |