CN109001143A - A kind of mid-infrared light spectrometry of sensitive prediction Chinese ephedra quality characteristic - Google Patents
A kind of mid-infrared light spectrometry of sensitive prediction Chinese ephedra quality characteristic Download PDFInfo
- Publication number
- CN109001143A CN109001143A CN201810913774.1A CN201810913774A CN109001143A CN 109001143 A CN109001143 A CN 109001143A CN 201810913774 A CN201810913774 A CN 201810913774A CN 109001143 A CN109001143 A CN 109001143A
- Authority
- CN
- China
- Prior art keywords
- ephedra
- chinese ephedra
- sample
- model
- prediction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 241001465251 Ephedra sinica Species 0.000 title claims abstract description 290
- 238000004611 spectroscopical analysis Methods 0.000 title claims abstract description 82
- 238000001228 spectrum Methods 0.000 claims abstract description 116
- 230000003595 spectral effect Effects 0.000 claims abstract description 102
- 238000000034 method Methods 0.000 claims abstract description 73
- 238000012360 testing method Methods 0.000 claims abstract description 53
- 238000005259 measurement Methods 0.000 claims abstract description 51
- 238000004519 manufacturing process Methods 0.000 claims abstract description 50
- 229930000044 secondary metabolite Natural products 0.000 claims abstract description 50
- FMCGSUUBYTWNDP-ONGXEEELSA-N (1R,2S)-2-(dimethylamino)-1-phenyl-1-propanol Chemical compound CN(C)[C@@H](C)[C@H](O)C1=CC=CC=C1 FMCGSUUBYTWNDP-ONGXEEELSA-N 0.000 claims abstract description 36
- FMCGSUUBYTWNDP-UHFFFAOYSA-N N-Methylephedrine Natural products CN(C)C(C)C(O)C1=CC=CC=C1 FMCGSUUBYTWNDP-UHFFFAOYSA-N 0.000 claims abstract description 36
- 229960002221 methylephedrine Drugs 0.000 claims abstract description 36
- 238000004433 infrared transmission spectrum Methods 0.000 claims abstract description 21
- 238000004458 analytical method Methods 0.000 claims abstract description 16
- 238000013528 artificial neural network Methods 0.000 claims abstract description 6
- 238000010238 partial least squares regression Methods 0.000 claims abstract description 3
- KWGRBVOPPLSCSI-WPRPVWTQSA-N (-)-ephedrine Chemical compound CN[C@@H](C)[C@H](O)C1=CC=CC=C1 KWGRBVOPPLSCSI-WPRPVWTQSA-N 0.000 claims description 60
- KWGRBVOPPLSCSI-UHFFFAOYSA-N d-ephedrine Natural products CNC(C)C(O)C1=CC=CC=C1 KWGRBVOPPLSCSI-UHFFFAOYSA-N 0.000 claims description 59
- 241000218671 Ephedra Species 0.000 claims description 58
- 230000009467 reduction Effects 0.000 claims description 37
- IOLCXVTUBQKXJR-UHFFFAOYSA-M potassium bromide Chemical compound [K+].[Br-] IOLCXVTUBQKXJR-UHFFFAOYSA-M 0.000 claims description 34
- 229960002179 ephedrine Drugs 0.000 claims description 30
- 108090000623 proteins and genes Proteins 0.000 claims description 30
- KWGRBVOPPLSCSI-WCBMZHEXSA-N pseudoephedrine Chemical compound CN[C@@H](C)[C@@H](O)C1=CC=CC=C1 KWGRBVOPPLSCSI-WCBMZHEXSA-N 0.000 claims description 29
- 229960003908 pseudoephedrine Drugs 0.000 claims description 24
- 241000723281 Ephedra equisetina Species 0.000 claims description 11
- 238000003705 background correction Methods 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 8
- 238000007873 sieving Methods 0.000 claims description 8
- 239000000835 fiber Substances 0.000 claims description 7
- 230000009466 transformation Effects 0.000 claims description 4
- 238000012937 correction Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 241000723237 Ephedra intermedia Species 0.000 claims description 2
- 239000003513 alkali Substances 0.000 claims description 2
- 238000000513 principal component analysis Methods 0.000 claims description 2
- 239000000126 substance Substances 0.000 abstract description 2
- 238000005457 optimization Methods 0.000 description 20
- 238000002360 preparation method Methods 0.000 description 20
- 238000002329 infrared spectrum Methods 0.000 description 12
- 230000035945 sensitivity Effects 0.000 description 11
- 241000196324 Embryophyta Species 0.000 description 9
- 239000000203 mixture Substances 0.000 description 7
- 238000010521 absorption reaction Methods 0.000 description 6
- 238000000605 extraction Methods 0.000 description 6
- 238000004566 IR spectroscopy Methods 0.000 description 5
- 238000004128 high performance liquid chromatography Methods 0.000 description 5
- 238000002203 pretreatment Methods 0.000 description 5
- 230000005540 biological transmission Effects 0.000 description 4
- 238000003113 dilution method Methods 0.000 description 4
- 238000010183 spectrum analysis Methods 0.000 description 4
- 238000002790 cross-validation Methods 0.000 description 3
- 238000004007 reversed phase HPLC Methods 0.000 description 3
- 244000025254 Cannabis sativa Species 0.000 description 2
- 238000004497 NIR spectroscopy Methods 0.000 description 2
- 238000003556 assay Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 241000411851 herbal medicine Species 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 239000007788 liquid Substances 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000002503 metabolic effect Effects 0.000 description 2
- 239000002207 metabolite Substances 0.000 description 2
- 238000002156 mixing Methods 0.000 description 2
- 239000000843 powder Substances 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000033764 rhythmic process Effects 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 235000015096 spirit Nutrition 0.000 description 2
- KSDSYIXRWHRPMN-UHFFFAOYSA-N 4'-O-beta-D-Galactopyranoside-6''-p-Coumaroylprunin-4',5,7-Trihydroxyflavanone Natural products OC1C(O)C(O)C(CO)OC1OC1=CC=C(C2OC3=CC(O)=CC(O)=C3C(=O)C2)C=C1 KSDSYIXRWHRPMN-UHFFFAOYSA-N 0.000 description 1
- 241000195955 Equisetum hyemale Species 0.000 description 1
- 238000000674 Fourier transform near infrared absorbance spectroscopy Methods 0.000 description 1
- 241000246500 Gentianella aspera Species 0.000 description 1
- VTAJIXDZFCRWBR-UHFFFAOYSA-N Licoricesaponin B2 Natural products C1C(C2C(C3(CCC4(C)CCC(C)(CC4C3=CC2)C(O)=O)C)(C)CC2)(C)C2C(C)(C)CC1OC1OC(C(O)=O)C(O)C(O)C1OC1OC(C(O)=O)C(O)C(O)C1O VTAJIXDZFCRWBR-UHFFFAOYSA-N 0.000 description 1
- DEMKZLAVQYISIA-ONJCETCRSA-N Liquiritin Natural products O([C@H]1[C@H](O)[C@@H](O)[C@H](O)[C@@H](CO)O1)c1ccc([C@@H]2Oc3c(C(=O)C2)ccc(O)c3)cc1 DEMKZLAVQYISIA-ONJCETCRSA-N 0.000 description 1
- DEMKZLAVQYISIA-UHFFFAOYSA-N Liquirtin Natural products OC1C(O)C(O)C(CO)OC1OC1=CC=C(C2OC3=CC(O)=CC=C3C(=O)C2)C=C1 DEMKZLAVQYISIA-UHFFFAOYSA-N 0.000 description 1
- 239000004793 Polystyrene Substances 0.000 description 1
- VYPSYNLAJGMNEJ-UHFFFAOYSA-N Silicium dioxide Chemical compound O=[Si]=O VYPSYNLAJGMNEJ-UHFFFAOYSA-N 0.000 description 1
- 229930013930 alkaloid Natural products 0.000 description 1
- 150000003797 alkaloid derivatives Chemical class 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 239000003153 chemical reaction reagent Substances 0.000 description 1
- 230000002060 circadian Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 229940079593 drug Drugs 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 238000001035 drying Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- LPLVUJXQOOQHMX-UHFFFAOYSA-N glycyrrhetinic acid glycoside Natural products C1CC(C2C(C3(CCC4(C)CCC(C)(CC4C3=CC2=O)C(O)=O)C)(C)CC2)(C)C2C(C)(C)C1OC1OC(C(O)=O)C(O)C(O)C1OC1OC(C(O)=O)C(O)C(O)C1O LPLVUJXQOOQHMX-UHFFFAOYSA-N 0.000 description 1
- 229960004949 glycyrrhizic acid Drugs 0.000 description 1
- UYRUBYNTXSDKQT-UHFFFAOYSA-N glycyrrhizic acid Natural products CC1(C)C(CCC2(C)C1CCC3(C)C2C(=O)C=C4C5CC(C)(CCC5(C)CCC34C)C(=O)O)OC6OC(C(O)C(O)C6OC7OC(O)C(O)C(O)C7C(=O)O)C(=O)O UYRUBYNTXSDKQT-UHFFFAOYSA-N 0.000 description 1
- 239000001685 glycyrrhizic acid Substances 0.000 description 1
- 235000019410 glycyrrhizin Nutrition 0.000 description 1
- LPLVUJXQOOQHMX-QWBHMCJMSA-N glycyrrhizinic acid Chemical compound O([C@@H]1[C@@H](O)[C@H](O)[C@H](O[C@@H]1O[C@@H]1C([C@H]2[C@]([C@@H]3[C@@]([C@@]4(CC[C@@]5(C)CC[C@@](C)(C[C@H]5C4=CC3=O)C(O)=O)C)(C)CC2)(C)CC1)(C)C)C(O)=O)[C@@H]1O[C@H](C(O)=O)[C@@H](O)[C@H](O)[C@H]1O LPLVUJXQOOQHMX-QWBHMCJMSA-N 0.000 description 1
- 238000003306 harvesting Methods 0.000 description 1
- 239000004615 ingredient Substances 0.000 description 1
- DEMKZLAVQYISIA-ZRWXNEIDSA-N liquiritin Chemical compound O[C@@H]1[C@@H](O)[C@H](O)[C@@H](CO)O[C@H]1OC1=CC=C([C@H]2OC3=CC(O)=CC=C3C(=O)C2)C=C1 DEMKZLAVQYISIA-ZRWXNEIDSA-N 0.000 description 1
- 239000004570 mortar (masonry) Substances 0.000 description 1
- 238000000491 multivariate analysis Methods 0.000 description 1
- 239000003960 organic solvent Substances 0.000 description 1
- 239000008188 pellet Substances 0.000 description 1
- 238000005220 pharmaceutical analysis Methods 0.000 description 1
- 229920002223 polystyrene Polymers 0.000 description 1
- 238000000746 purification Methods 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 239000012898 sample dilution Substances 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000024053 secondary metabolic process Effects 0.000 description 1
- 239000002689 soil Substances 0.000 description 1
- 239000000243 solution Substances 0.000 description 1
- GSZUGBAEBARHAW-UHFFFAOYSA-N sophoraflavone B Natural products OC1C(O)C(O)C(CO)OC1OC1=CC=C(C=2OC3=CC(O)=CC=C3C(=O)C=2)C=C1 GSZUGBAEBARHAW-UHFFFAOYSA-N 0.000 description 1
- 238000007811 spectroscopic assay Methods 0.000 description 1
- 239000010902 straw Substances 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 238000012795 verification Methods 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
- G01N21/3563—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light for analysing solids; Preparation of samples therefor
-
- 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
- G01N21/3563—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light for analysing solids; Preparation of samples therefor
- G01N2021/3568—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light for analysing solids; Preparation of samples therefor applied to semiconductors, e.g. Silicon
Landscapes
- Physics & Mathematics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The invention discloses a kind of mid-infrared light spectrometries of sensitive prediction Chinese ephedra quality characteristic, and the quality characteristic is type, the place of production, in a few days plucking time and secondary metabolites content are any one or more of；It first collects the Chinese ephedra plant sample of different quality characteristics and is prepared into test sample, infrared transmission spectra in measurement, gained spectrum is pre-processed without or through Chemical Measurement, selection modeling spectral region and characteristic variable, using techniques of discriminant analysis or it is opposite propagate artificial neural network and establish respectively predict the qualitative model of Chinese ephedra type, the place of production, in a few days plucking time, the quantitative model that prediction Chinese ephedra secondary metabolites content is established using Partial Least Squares Regression or back-propagation artificial neural network, finally predicts the quality characteristic of unknown Chinese ephedra sample using model built；Compared with prior art, the method for the present invention improves the right judging rate for the in a few days plucking time for being difficult to differentiate, and realizes the Accurate Prediction of low content secondary metabolite methylephedrine.
Description
Technical field
The present invention relates to the prediction techniques of plant quality characteristic, use middle infrared spectrum skill more specifically to a kind of
The method of the sensitive prediction Chinese ephedra quality characteristic of art.
Background technique
Human use Chinese ephedra and its secondary metabolite it is with a long history, it has been recognized that the quality characteristic of Chinese ephedra by
The influence of the factors such as heredity, ecological environment and metabolic rhythm.In other words, the type of Chinese ephedra, the place of production or/and plucking time be not
Together, the composition of secondary metabolite or/and content are different, and the value of Chinese ephedra is also just different.With resources such as plant, soil
Scarcity is increasingly sharpened, and the utilization rate for improving resource becomes the task of top priority.So establishing the side of sensitive prediction Chinese ephedra quality characteristic
Method has important meaning for the resource of the changing rule of research Chinese ephedra quality, scientific utilization Chinese ephedra and its secondary metabolite
Justice.
Currently, the evaluation method of published Chinese ephedra quality characteristic mainly includes the micro- mirror based on micromorphology information
Other method (" Chinese Pharmacopoeia " version in 2015) and high performance liquid chromatography (HPLC) method (" Chinese Pharmacopoeia " 2015 based on chemical information
Version).Although these analysis methods achieve preferable effect in terms of the qualitative and quantitatively characterizing of Chinese ephedra, they
In the presence of more subjective or sample pre-treatments complexity, sample broke, time-consuming, organic solvent pollution environment, it is difficult to realize on-line checking
The disadvantages of.
Near infrared spectrum (NIR) method has been used for type, the place of production, plucking time and the alkaloid of predicting Chinese ephedra plant
(Fan Q, Wang Y, Sun P, et al, Discrimination of Ephedra plants with diffuse
Reflectance FT-NIRS and multivariate analysis [J] .Talanta, 2010,80 (3): 1245-
1250.；Yi Zhenkui, Fan Qi, Wang Liqiong wait near-infrared to diffuse spectrometry combination CP-ANN and PLS high throughput analysis ephedra sinica
Medicinal material [J] Pharmaceutical Analysis magazine, 2012 (8): 1402-1408.).Although near infrared spectroscopy overcomes chromatographic many scarce
Point, regrettably, near infrared spectrum are mainly that the frequency multiplication of hydric group vibration and sum of fundamental frequencies absorb, therefore absorption coefficient is small, detection
Sensitivity is low, be not suitable for the small quality characteristic of characterization difference (for example, to the differentiation of Chinese ephedra in a few days plucking time there are mistake,
Right judging rate is 93.3%) and the low secondary metabolite of content is (for example, methylephedrine to content lower than 0.1% is determined
Amount prediction does not obtain acceptable effect).It is therefore desirable to establish a kind of quality that the difference for capableing of sensitive prediction Chinese ephedra is small
The method of characteristic and the low secondary metabolite of content.
Middle infrared spectrum (IR) is mainly that the fundamental frequency of molecular vibration absorbs, and compared near infrared spectrum, absorption coefficient is big, inspection
Survey high sensitivity.Although type, the place of production, picking time and secondary metabolite that mid-infrared light spectrometry has been used for pre- measuring plants contain
Amount, regrettably, the presently disclosed prior art does not give full play to high sensitivity feature possessed by mid-infrared light spectrometry
To overcome the muting sensitivity of near infrared spectroscopy to limit to.For example, being needle when the picking time of existing mid-infrared light spectrometry prediction plant
(Shen rosy clouds, Zhao Yanli, Zhang Ji wait Different Harvesting Time Yunnan rough gentian to the plant harvested to the different months that quality characteristic differs greatly
Infrared spectroscopy identify research [J] spectroscopy and spectrum analysis, 2016,36 (5): 1358-1362.)；Predict Plant Secondary Materials generation
It is mostly that not less than 0.1% secondary metabolite, (Zhan Xueyan, Lin Zhaozhou, Sun Yang wait Radix Glycyrrhizae for content when thanking to product assay
The screening and parsing [J] spectroscopy and spectrum analysis of middle liquiritin and glycyrrhizic acid infrared quantitative aspect of model variable, 2015 (9):
2530-2535.).So far, there is not yet the quality characteristic for using the difference of the sensitive pre- measuring plants of mid-infrared light spectrometry small with
And the secondary metabolite that content is low.
Summary of the invention
It is an object of the invention to overcome the shortcomings of the prior art, the sensitive prediction Chinese ephedra quality characteristic of one kind is provided
Mid-infrared light spectrometry, to predict including small low secondary of quality characteristic (for example, in a few days plucking time) and content of difference
The a variety of quality characteristics of Chinese ephedra including metabolite (for example, methylephedrine).
Through studying, the invention provides the following technical scheme:
A kind of mid-infrared light spectrometry of sensitive prediction Chinese ephedra quality characteristic, the quality characteristic are type, the place of production, in a few days adopt
It plucks the time and secondary metabolites content is any one or more of；Wherein,
Predict Chinese ephedra type mid-infrared light spectrometry the following steps are included:
(1) it collects and records different types of Chinese ephedra plant sample, it is dry, its herbaceous stem stem is taken, Chinese ephedra sample is obtained；The fiber crops
The type of yellow plant is ephedra sinica (Ephedra sinica Stapf), epheday intermedia (Ephedra intermedia Schrenk
Et C.A.Mey.) or ephedra equisetina (Ephedra equisetina Bge.)；
(2) by Chinese ephedra sample comminution, sieving obtains Ephedra sample；
(3) Ephedra sample is mixed with potassium bromide, suppresses, obtains Ephedra kbr tablet, as test sample, use is red
External spectrum instrument is in 4000~400cm-1The middle infrared transmission spectra of measurement test sample in range, every time with identical before scanning test sample
Parameter scanning and background correction；
(4) gained spectrum is not preprocessed or is smoothly SGS pretreatment through Savitzky-Golay, chooses modeling spectrum model
The wave number upper limit value enclosed is 3936 ± 64cm-1That is 4000~3872cm-1, wave number lower limit value be 464 ± 64cm-1I.e. 528~
400cm-1, using Principal Component Analysis, that is, PCA dimensionality reductions, preceding 8 principal components are chosen according to the sequence of contribution rate from high to low, are established
Predict discriminant analysis, that is, DA model of Chinese ephedra type；
Alternatively, gained spectrum chooses identical modeling spectral region and identical master after first derivative, that is, FD pretreatment
Ingredient establishes the opposite of prediction Chinese ephedra type and propagates artificial neural network, that is, CP-ANN model；
(5) the Chinese ephedra sample for taking unknown type, by step (2), (3) and (4) the method prepare test sample, in measurement it is red
Outer transmitted spectrum simultaneously carries out spectroscopic data processing, and then applying step (4) model built predicts the type of unknown Chinese ephedra sample；
Predict the Chinese ephedra place of production mid-infrared light spectrometry the following steps are included:
(1) the same type Chinese ephedra plant sample of different sources is collected and records, it is dry, its herbaceous stem stem is taken, Chinese ephedra sample is obtained；
The place of production of the Chinese ephedra plant is Shanxi or the Inner Mongol；
(2) by Chinese ephedra sample comminution, sieving obtains Ephedra sample；
(3) Ephedra sample is mixed with potassium bromide, suppresses, obtains Ephedra kbr tablet, as test sample, use is red
External spectrum instrument is in 4000~400cm-1The middle infrared transmission spectra of measurement test sample in range, every time with identical before scanning test sample
Parameter scanning and background correction；
(4) gained spectrum is not preprocessed or pre-processes through SGS, and the wave number upper limit value for choosing modeling spectral region is 3819
±181cm-1That is 4000~3638cm-1, wave number lower limit value be 581 ± 181cm-1That is 762~400cm-1, using PCA dimensionality reduction, press
Preceding 6 principal components are chosen according to the sequence of contribution rate from high to low, establish the DA model in the prediction Chinese ephedra place of production；
Alternatively, gained spectrum is not preprocessed, identical modeling spectral region and identical principal component are chosen, establishes prediction
The CP-ANN model in the Chinese ephedra place of production；
(5) the Chinese ephedra sample for taking the unknown place of production, by step (2), (3) and (4) the method prepare test sample, in measurement it is red
Outer transmitted spectrum simultaneously carries out spectroscopic data processing, and then applying step (4) model built predicts the place of production of unknown Chinese ephedra sample；
Predict Chinese ephedra in a few days plucking time mid-infrared light spectrometry the following steps are included:
(1) the similar place of production Chinese ephedra plant sample of the same race of different in a few days plucking times is collected and records, it is dry, take its herbaceous stem
Stem obtains Chinese ephedra sample；The plucking time of the Chinese ephedra plant is the morning or afternoon of same day；
(2) by Chinese ephedra sample comminution, sieving obtains Ephedra sample；
(3) Ephedra sample is mixed with potassium bromide, suppresses, obtains Ephedra kbr tablet, as test sample, use is red
External spectrum instrument is in 4000~400cm-1The middle infrared transmission spectra of measurement test sample in range, every time with identical before scanning test sample
Parameter scanning and background correction；
(4) gained spectrum is SNV pretreatment through multiplicative scatter correction, that is, MSC or standard normal transformation, chooses modeling spectrum
The wave number upper limit value of range is 3815 ± 117cm-1That is 3932~3698cm-1, wave number lower limit value be 585 ± 117cm-1I.e. 702~
468cm-1, using PCA dimensionality reduction, preceding 4 principal components are chosen according to the sequence of contribution rate from high to low, prediction Chinese ephedra is established and in a few days adopts
Pluck the DA model of time；
Alternatively, gained spectrum is pre-processed through MSC, identical modeling spectral region and identical principal component are chosen, is established pre-
Survey the CP-ANN model of Chinese ephedra in a few days plucking time；
(5) the Chinese ephedra sample for taking plucking time in unknown day, by step (2), (3) and (4) the method prepare test sample,
Infrared transmission spectra and spectroscopic data processing is carried out in measurement, then applying step (4) model built predicts unknown Chinese ephedra sample
In a few days plucking time；
Predict Chinese ephedra in secondary metabolites content mid-infrared light spectrometry the following steps are included:
(1) the same type Chinese ephedra plant sample of different secondary metabolites contents is collected and records, it is dry, take its herbaceous stem
Stem obtains Chinese ephedra sample；The secondary metabolite is that ephedrine, d-pseudo-ephedrine and methylephedrine are any one or more of；
(2) by Chinese ephedra sample comminution, sieving obtains Ephedra sample；Measure secondary metabolite in Ephedra sample
The reference value of content；
(3) Ephedra sample is mixed with potassium bromide, suppresses, obtains Ephedra kbr tablet, as test sample, use is red
External spectrum instrument is in 4000~400cm-1The middle infrared transmission spectra of measurement test sample in range, every time with identical before scanning test sample
Parameter scanning and background correction；
(4) when establishing the prediction model of Chinese ephedra Ephedrine content, gained spectrum is through second dervative, that is, SD and mean value center
Changing is MC pretreatment, and the wave number upper limit value for choosing modeling spectral region is 3986 ± 14cm-1That is 4000~3972cm-1, under wave number
Limit value is 414 ± 14cm-1That is 428~400cm-1, using PCA dimensionality reduction, first 4 are chosen according to the sequence of contribution rate from high to low
Main gene, the Partial Least Squares Regression, that is, PLS or back-propagation artificial neural network for establishing prediction Chinese ephedra Ephedrine content are
BP-ANN model；
When establishing the prediction model of Determination of Pseudoephedrine in Chinese ephedra, gained spectrum is pre-processed through SD and MC, chooses modeling light
The wave number upper limit value of spectral limit is 3930 ± 70cm-1That is 4000~3860cm-1, wave number lower limit value be 470 ± 70cm-1I.e. 540~
400cm-1, using PCA dimensionality reduction, preceding 4 main genes are chosen according to the sequence of contribution rate from high to low, establish pseudo- fiber crops in prediction Chinese ephedra
PLS the or BP-ANN model of yellow alkali content；
When establishing the prediction model of methylephedrine content in Chinese ephedra, gained spectrum is pre-processed through FD and MC, chooses modeling
The wave number upper limit value of spectral region is 3800 ± 200cm-1That is 4000~3600cm-1, wave number lower limit value be 600 ± 200cm-1I.e.
800~400cm-1, using PCA dimensionality reductions, preceding 4 main genes are chosen according to the sequence of contribution rate from high to low, establish prediction Chinese ephedra
PLS the or BP-ANN model of middle methylephedrine content；
(5) the Chinese ephedra sample for taking unknown secondary metabolites content is prepared by step (2), (3) and (4) the method and is supplied
Infrared transmission spectra and spectroscopic data processing is carried out in test product, measurement, then applying step (4) model built predicts unknown Chinese ephedra
The secondary metabolites content of sample.
The method of the present invention can individually or simultaneously it is sensitive prediction Chinese ephedra a variety of quality characteristics: type (hereditary capacity), produce
Ground (ecological environment characteristic), in a few days plucking time (metabolic rhythm characteristic) and secondary metabolites content.Different types of Chinese ephedra
Plant causes the composition of its secondary metabolite or/and content different due to inherent cause difference.The method of the present invention is based on secondary
The composition of raw metabolite or/and the difference of content be suitable for predicting different types of Chinese ephedra, for example, ephedra sinica, epheday intermedia and
Ephedra equisetina etc..With the Chinese ephedra plant of type different sources, although inherent cause is identical, influenced by different ecological environment
Its growth and development shows difference to a certain extent, so as to cause its secondary metabolite composition or/and content certain
It is different in degree.The difference of composition or/and content of the method for the present invention based on secondary metabolite is suitable for predicting with type not
With the Chinese ephedra in the place of production, for example, Inner Mongol ephedra sinica and Shanxi ephedra sinica etc..The similar place of production of the same race in a few days different time (morning with
Afternoon) picking Chinese ephedra plant, although inherent cause and ecological environment are all identical, by circadian influenced its growth send out
Educate and show difference in lesser degree, so as to cause its secondary metabolite composition or/and content in lesser degree not
Together.The difference of composition or/and content of the method for the present invention based on secondary metabolite is suitable for predicting the similar place of production of the same race in a few days
The Chinese ephedra of different time picking, for example, the Shanxi ephedra sinica etc. in same morning day and picking in afternoon.
The mid-infrared light spectrometry of the prediction Chinese ephedra type, the place of production, in a few days plucking time or/and secondary metabolites content
The step of (1) in, the Chinese ephedra plant sample of collection is dried, its object is to make Chinese ephedra plant sample contained humidity reduction
It is able to maintain metastable level to its quality characteristic, dry method can be using relatively wet in room temperature and 40%~60%
It is placed under the conditions of degree, or does not influence the proper method of Chinese ephedra plant sample quality characteristic itself using other, such as low temperature drying,
Sunning etc..
The mid-infrared light spectrometry of the prediction Chinese ephedra type, the place of production, in a few days plucking time or/and secondary metabolites content
The step of (2) in, Chinese ephedra sample is crushed, is sieved, its object is to obtain the suitable uniform powder of granularity in favor of pressure
Ephedra kbr tablet processed and measurement Chinese ephedra secondary metabolite (ephedrine, d-pseudo-ephedrine, methylephedrine) content.This hair
The method of bright preferred Ephedra granularity, that is, mesh number is to measure different meshes (≤20,20~40,40~60,60 using HPLC method
~80,80~100,100~200,200 mesh of >) Ephedra sample secondary metabolite (ephedrine, d-pseudo-ephedrine, methyl
Ephedrine) content, as a result, it has been found that, when the mesh number of Ephedra is 100~200 mesh, content height, the powder of secondary metabolite
The amount of sample is more and easily mixes with potassium bromide tabletted.Therefore, grit number is crossed in the step (2) is preferably 100~200
Mesh.
Obtaining the distortionless highly sensitive spectrum of sample message is the basis for establishing high-quality prediction model.Sample pre-treatments letter
Single, the analysis application and popularization easy to operate for being conducive to analysis method.The prediction Chinese ephedra type, the place of production, in a few days plucking time
Or/and secondary metabolites content mid-infrared light spectrometry the step of (3) in, be that Ephedra is mixed and suppressed with potassium bromide
At Ephedra kbr tablet as test sample, wherein infrared transmission spectra is then measured.The present invention is first to spectral measurement mould
Formula has carried out preferably, comparing diffusing reflection measurement pattern and transmission measurement mode, does not pollute although diffusing reflection measurement pattern has
Ephedra sample does not need the advantages of chemical reagent, but since absorption signal is too strong, or even hypersorption occurs, so selection
The transmission measurement mode of absorption is reduced by dilute sample.Secondly, test sample dilution side of the present invention to transmission measurement mode
Method has carried out preferably, comparing liquid dilution method and solid dilution method, due to liquid dilution method need to Ephedra sample into
Row is complicated, time-consuming pre-treatment, so having selected to hardly need the solid dilution method i.e. pellet technique of sample pre-treatments.
When being diluted using potassium bromide, it is contemplated that absorption signal too strong (being higher than upper limit of quantification) or too weak (lower than inspection
Survey limit/lower limit of quantitation) be all unfavorable for spectrum analysis and Accurate Prediction, the present invention to the mixed proportion of Ephedra and potassium bromide into
It has gone preferably, as a result, it has been found that, when the mixing mass ratio of Ephedra and potassium bromide is 1: 150, absorption signal is moderate, is conducive to light
Spectrum analysis and Accurate Prediction.Therefore, the prediction Chinese ephedra type, the place of production, in a few days plucking time or/and secondary metabolites content
Mid-infrared light spectrometry the step of (3) in, the mixing mass ratio of Ephedra and potassium bromide is preferably 1: 150.
When suppressing Ephedra kbr tablet, it is contemplated that integrality, translucency of the piece etc. have an impact to spectrum, this hair
It is bright that tableting pressure and tabletting time have been carried out preferably, as a result, it has been found that, when tableting pressure is 10~30MPa, the tabletting time be 1~
At 2 minutes, the Ephedra kbr tablet quality of compacting is best.Therefore, the prediction Chinese ephedra type, the place of production, when in a few days picking
Between or/and secondary metabolites content mid-infrared light spectrometry the step of (3) in, the tableting pressure of Ephedra kbr tablet is excellent
It is selected as 10~30MPa, the tabletting time is preferably 1~2 minute.
The present invention has further carried out preferably the resolution ratio of spectral measurement and scanning times.Scanning constant number is 16
It is secondary, it is respectively 2cm with resolution ratio-1、4cm-1、8cm-1、16cm-1、32cm-1Same test sample is measured in parallel 5 times, synthesis variance
Size and variance spectrum smoothness obtain resolution ratio be 4cm-1Shi Guangpu is best, and therefore, the prediction Chinese ephedra type produces
In the step of mid-infrared light spectrometry of ground, in a few days plucking time or/and secondary metabolites content (3), the resolution of spectral measurement
Rate is preferably 4cm-1.Fixed resolution is 4cm-1, it is respectively 16,32,64,128 times parallel to same test sample with scanning times
Measurement 5 times, as a result, it has been found that, spectral variance is larger when scanning times are 16 times, variance when scanning times are 32 times, 64 times, 128 times
Spectrum no significant difference, and scanning times are more, spent time is longer, and therefore, the prediction Chinese ephedra type, is in a few days adopted at the place of production
In the step of plucking the mid-infrared light spectrometry of time or/and secondary metabolites content (3), the scanning times of spectral measurement are preferably
Not less than 32 times.
After obtaining the distortionless highly sensitive spectrum of sample message, in order to improve specificity, sensitivity and the accuracy of analysis,
Need further creatively to excavate the effective information of target signature in spectroscopic data using chemometric techniques combination,
In, the selection of Pretreated spectra scheme, modeling spectral region and modeling variable (principal component or main gene) is the technology of the present invention side
The key point and difficult point of case.The signal-to-noise ratio that can be improved spectrum using Pretreated spectra scheme appropriate is conducive to improve model
Estimated performance.But if Pretreated spectra scheme is inappropriate, then run counter to desire, reduces the signal-to-noise ratio of spectrum.Using appropriate
Modeling spectral region and modeling variable the effective information of target signature can be extracted from complicated spectroscopic data, make to be modeled
Type has good estimated performance.But if selected modeling spectral region includes too many redundancy, the simplification of spectral information
And clean lack, the estimated performance decline of model built；If the simplification and purification of spectral information are excessively, mesh in spectroscopic data
The effective information for marking feature retains deficiency, and the estimated performance of model built also declines.Similarly, if selected modeling variable too
More or very little, the estimated performance of model all declines.The prediction Chinese ephedra type, the place of production, in a few days plucking time or/and secondary metabolism
In the step of mid-infrared light spectrometry of product assay (4), selected Pretreated spectra scheme, modeling spectral region and modeling become
Amount is all creative work achievement of the invention, is the specificity for guaranteeing analysis method of the present invention, sensitivity and the pass of accuracy
Key.
The beneficial effects of the present invention are: the method for the present invention is formed based on secondary metabolite or/and the difference of content, point
It does not measure infrared in the Chinese ephedra sample of different quality characteristics (type, the place of production, in a few days plucking time, secondary metabolites content)
Spectrum establishes the qualitative model of prediction Chinese ephedra type, the place of production, in a few days plucking time in conjunction with chemometric techniques respectively, with
And the quantitative model of prediction Chinese ephedra secondary metabolite (ephedrine, d-pseudo-ephedrine, methylephedrine) content, it can individually or together
Shi Lingmin predicts a variety of quality characteristics (type, the place of production, in a few days plucking time, secondary metabolites content) of Chinese ephedra, except having
Outside the advantages that sample pre-treatments are simple, analysis is easy to operate, it is often more important that, compared with prior art, improves and be difficult to differentiate
Chinese ephedra in a few days plucking time correct decision rate, and realize the standard of low content secondary metabolite methylephedrine in Chinese ephedra
True quantitative forecast.
Detailed description of the invention
Fig. 1 is the DA figure for predicting Chinese ephedra type: capitalization A, B, C respectively represent ephedra sinica, epheday intermedia and ephedra equisetina
Calibration set sample, lowercase a, b, c respectively represent the verifying collection sample of ephedra sinica, epheday intermedia and ephedra equisetina.
Fig. 2 is the CP-ANN figure for predicting Chinese ephedra type: white area represents ephedra sinica, and gray area represents epheday intermedia, deep
Gray area represents ephedra equisetina；Capitalization A, B, C respectively represent the calibration set sample of ephedra sinica, epheday intermedia and ephedra equisetina
Product, lowercase a, b, c respectively represent the verifying collection sample of ephedra sinica, epheday intermedia and ephedra equisetina.
Fig. 3 is the DA figure for predicting the Chinese ephedra place of production: capitalization A and B respectively represent the correction in Shanxi and Inner Mongol ephedra sinica
Collect sample, lowercase a and b respectively represent the verifying collection sample of Shanxi and Inner Mongol ephedra sinica.
Fig. 4 is the CP-ANN figure for predicting the Chinese ephedra place of production: white area represents Shanxi ephedra sinica, and gray area represents the Inner Mongol
Ephedra sinica；Capitalization A and B respectively represent the calibration set sample of Shanxi and Inner Mongol ephedra sinica, lowercase a and b generation respectively
The verifying collection sample in table Shanxi and Inner Mongol ephedra sinica.
Fig. 5 be predict Chinese ephedra in a few days plucking time DA figure: capitalization A and B respectively represent the morning picking and afternoon adopt
The calibration set sample of Shanxi ephedra sinica is plucked, lowercase a and b respectively represent morning picking and pick testing for Shanxi ephedra sinica in the afternoon
Card collection sample.
Fig. 6 be predict Chinese ephedra in a few days plucking time CP-ANN figure: white area represent the morning picking Shanxi ephedra sinica,
Gray area represents the Shanxi ephedra sinica of picking in afternoon；Capitalization A and B respectively represent morning picking and picking in afternoon Shanxi grass
The calibration set sample of Chinese ephedra, lowercase a and b respectively represent the verifying collection sample of morning picking and picking in afternoon Shanxi ephedra sinica
Product.
Fig. 7 is that Chinese ephedra Ephedrine content PLS model reference value and the linearly related of predicted value are schemed.
Fig. 8 is that Chinese ephedra Ephedrine content BP-ANN model reference value and the linearly related of predicted value are schemed.
Fig. 9 is that Determination of Pseudoephedrine PLS model reference value and the linearly related of predicted value are schemed in Chinese ephedra.
Figure 10 is that Determination of Pseudoephedrine BP-ANN model reference value and the linearly related of predicted value are schemed in Chinese ephedra.
Figure 11 is that methylephedrine content PLS model reference value and the linearly related of predicted value are schemed in Chinese ephedra.
Figure 12 is that methylephedrine content BP-ANN model reference value and the linearly related of predicted value are schemed in Chinese ephedra.
Specific embodiment
To make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawing to of the invention preferred
Embodiment is described in detail.
Instrument involved in embodiment: Spectrum One Fourier transformation infrared spectrometer (U.S. Perkin Elmer
Company), equipped with DTGS KBr detector and transmission sampling attachment；YP-2 tablet press machine (Shanghai lofty mountains scientific instrument Co., Ltd).
A kind of middle infrared transmission spectra method of sensitive prediction Chinese ephedra type of embodiment 1
1. the collection of Chinese ephedra plant sample
Collect and record different types of Chinese ephedra plant sample, including 12 batches of ephedra sinica, 3 batches of epheday intermedia and 1 batch of scouring rush fiber crops
Huang, placement is reduced to Chinese ephedra plant sample contained humidity metastable under room temperature and 40%~60% relative humidities
Level takes its herbaceous stem stem, obtains Chinese ephedra sample.
2. the preparation of Ephedra sample
It is crushed under the power of 250W 50 seconds (every crush 10 seconds is spaced 2 seconds), crosses 100~200 meshes, obtain Ephedra sample
Product.
3. the preparation of test sample and the measurement of middle infrared spectrum
Ephedra sample and potassium bromide (spectroscopic pure) are taken, in mass ratio 1: 150 grinds well in the agate mortar, through tablet press machine
It is pushed 1 minute in 20MPa, Ephedra kbr tablet is made, as test sample.
After Fourier transformation infrared spectrometer is fully warmed-up and passes through verification, the polystyrene film being equipped at random is used
Confirm that it meets spectroscopic assay requirement.With resolution ratio 4cm-1With scanning times 32 times in 4000~400cm-1Measurement supplies in range
The middle infrared transmission spectra of test product, every time with identical parameters scanning and background correction before scanning test sample.
4. the extraction and modeling of spectral signature variable
Choose 9 batches of ephedra sinica, 2 batches of epheday intermedia and 1 batch respectively from 12 batches of ephedra sinica, 3 batches of epheday intermedia and 1 batch of ephedra equisetina
Ephedra equisetina is as calibration set sample, remaining is as verifying collection sample.
The selection and estimated performance of the main modeling parameters of 4.1DA model
4.1.1 the selection of Pretreated spectra scheme
When establishing the DA model of prediction Chinese ephedra type, in order to obtain the optimum prediction performance of model, to without locating in advance
Reason is that NP, MSC, SNV, FD, SD, SGS, Norris are smoothly that the multiple spectrums preconditioning techniques such as NDS are screened and combined,
It is shown in Table 1.The result shows that the estimated performance of built DA model is optimal when gained spectrum is pre-processed through NP or SGS.4.1.2 light is modeled
The selection of spectral limit
It is automatic in modeling software to be modeled according to tested characteristic using aforementioned preferred Pretreated spectra scheme for spectral region
Artificial optimization is 3936 ± 64cm of wave number upper limit value on the basis of selection-1That is 4000~3872cm-1, wave number lower limit value 464 ±
64cm-1That is 528~400cm-1When, the calibration set right judging rate and verifying collection right judging rate of model can reach 100.0%, so excellent
Changing the spectral region in range can be used to model, such as the model 1,13,14 and 15 in table 1；And work as modeling spectral region not
When in optimization range, the calibration set right judging rate or/and verifying collection right judging rate of model are not up to 100.0%, such as the model in table 1
16。
4.1.3 the selection of the dimensionality reduction of spectroscopic data and modeling variable (principal component)
PCA dimensionality reduction is carried out to the spectroscopic data in selected modeling spectral region, when being modeled using different principal components, mould
The performance of type is there are notable difference, such as the model 1,11 and 12 in table 1, finally according to contribution rate sequential selection from high to low before
8 principal components predict the DA model of Chinese ephedra type to establish.
4.1.4 the foundation and verifying of model
The DA model with verifying prediction Chinese ephedra type is established using the principal component scores of calibration set and verifying collection spectroscopic data,
(model 1 in table 1) as shown in Figure 1, calibration set right judging rate and verifying collection right judging rate are 100.0%, illustrate that the model has
There is good estimated performance.
The main modeling parameters and estimated performance of 1 Chinese ephedra type DA model of table
The selection and estimated performance of the main modeling parameters of 4.2CP-ANN model
4.2.1 the selection of Pretreated spectra scheme
Establish prediction Chinese ephedra type CP-ANN model when, in order to obtain the optimum prediction performance of model, to NP, MSC,
The multiple spectrums preconditioning technique such as SNV, FD, SD, SGS, NDS is screened and has been combined, and is shown in Table 2.The result shows that gained spectrum
When pre-processing through FD, the estimated performance of built CP-ANN model is optimal.
4.2.2 the selection of spectral region is modeled
It is identical as DA model as used sample spectra when verifying CP-ANN model due to establishing, and two kinds of models
Prediction characteristic is similarly the type of Chinese ephedra, i.e., the effective information of same target feature is characterized in same spectra in same area, only
It is modeling algorithm difference, so CP-ANN model is identical as modeling spectral region used in DA model, i.e., the modeling light in table 2
3936~464cm of spectral limit-1It has been the modeling spectral region after optimization.
The main modeling parameters and estimated performance of 2 Chinese ephedra type CP-ANN model of table
4.2.3 the selection of the dimensionality reduction of spectroscopic data and modeling variable (principal component)
PCA dimensionality reduction is carried out to the spectroscopic data in selected modeling spectral region, when being modeled using different principal components, mould
The performance of type is there are notable difference, such as the model 4,11 and 12 in table 2, finally according to contribution rate sequential selection from high to low before
8 principal components predict the CP-ANN model of Chinese ephedra type to establish.
4.2.4 the foundation and verifying of model
The CP-ANN with verifying prediction Chinese ephedra type is established using the principal component scores of calibration set and verifying collection spectroscopic data
Model, (model 4 in table 2) as shown in Figure 2, calibration set right judging rate, cross validation right judging rate and verifying collection right judging rate are
100.0%, illustrate that the model has good estimated performance.
By Fig. 1 and 2 as it can be seen that linear DA model and non-linear CP-ANN model that the present embodiment is established can it is sensitive,
Accurately predict the type of Chinese ephedra, it was demonstrated that the highly sensitive spectrum of Chinese ephedra sample that the method for the present invention obtains contains the feature letter of sample
Breath.Although the original spectral data pretreating scheme that two kinds of models use is different, the modeling spectral region and modeling used becomes
Measure identical, and two kinds of models all have good estimated performance, confirmed each other the present invention accurately extracted from infrared spectroscopy and
The effective information of target signature is utilized, ensure that specificity, sensitivity and the accuracy of analysis method of the present invention.
A kind of middle infrared transmission spectra method in the sensitive prediction Chinese ephedra place of production of embodiment 2
The present embodiment uses preparation method of test article same as Example 1, spectral measurement condition and method, with embodiment
1 difference is that the present embodiment uses the Chinese ephedra plant sample of different sources to establish the pre- of the Chinese ephedra place of production with preferred modeling parameters
Survey model.
1. the collection of Chinese ephedra plant sample
Collect and record the ephedra sinica plant sample of different sources, including 24 batches of Shanxi ephedra sinica and 4 batches of Inner Mongol straws
Huang, placement is reduced to Chinese ephedra plant sample contained humidity metastable under room temperature and 40%~60% relative humidities
Level takes its herbaceous stem stem, obtains Chinese ephedra sample.
2. the preparation of Ephedra sample
With embodiment 1.
3. the preparation of test sample and the measurement of middle infrared spectrum
With embodiment 1.
4. the extraction and modeling of spectral signature variable
18 batches of Shanxi ephedra sinica and 3 batches of Inner Mongol are chosen respectively from 24 batches of Shanxi ephedra sinica and 4 batches of Inner Mongol ephedra sinica
Ephedra sinica is as calibration set sample, remaining is as verifying collection sample.
The selection and estimated performance of the main modeling parameters of 4.1DA model
4.1.1 the selection of Pretreated spectra scheme
Establish prediction the Chinese ephedra place of production DA model when, in order to obtain the optimum prediction performance of model, to NP, MSC, SNV,
The multiple spectrums preconditioning technique such as FD, SD, SGS, NDS is screened and has been combined, and is shown in Table 3.The result shows that gained spectrum is through NP
Or when SGS pretreatment, the estimated performance of built DA model is optimal.
4.1.2 the selection of spectral region is modeled
It is automatic in modeling software to be modeled according to tested characteristic using aforementioned preferred Pretreated spectra scheme for spectral region
Artificial optimization is 3819 ± 181cm of wave number upper limit value on the basis of selection-1That is 4000~3638cm-1, wave number lower limit value 581 ±
181cm-1That is 762~400cm-1When, the calibration set right judging rate and verifying collection right judging rate of model can reach 100.0%, so excellent
Changing the spectral region in range can be used to model, such as the model 1,13,14 and 15 in table 3；And work as modeling spectral region not
When in optimization range, the calibration set right judging rate or/and verifying collection right judging rate of model are not up to 100.0%, such as the model in table 3
16。
4.1.3 the selection of the dimensionality reduction of spectroscopic data and modeling variable (principal component)
PCA dimensionality reduction is carried out to the spectroscopic data in selected modeling spectral region, when being modeled using different principal components, mould
The performance of type is there are notable difference, such as the model 1,11 and 12 in table 3, finally according to contribution rate sequential selection from high to low before
6 principal components predict the DA model in the Chinese ephedra place of production to establish.
The main modeling parameters and estimated performance of 3 Chinese ephedra place of production DA model of table
4.1.4 the foundation and verifying of model
The DA model with the verifying prediction Chinese ephedra place of production is established using the principal component scores of calibration set and verifying collection spectroscopic data,
(model 1 in table 3) as shown in Figure 3, calibration set right judging rate and verifying collection right judging rate are 100.0%, illustrate that the model has
There is good estimated performance.
The selection and estimated performance of the main modeling parameters of 4.2CP-ANN model
4.2.1 the selection of Pretreated spectra scheme
Establish prediction the Chinese ephedra place of production CP-ANN model when, in order to obtain the optimum prediction performance of model, to NP, MSC,
The multiple spectrums preconditioning technique such as SNV, FD, SD, SGS, NDS is screened and has been combined, and is shown in Table 4.The result shows that gained spectrum
When pre-processing through NP, the estimated performance of built CP-ANN model is optimal.
4.2.2 the selection of spectral region is modeled
It is identical as DA model as used sample spectra when verifying CP-ANN model due to establishing, and two kinds of models
Prediction characteristic is similarly the place of production of Chinese ephedra, i.e., the effective information of same target feature is characterized in same spectra in same area, only
It is modeling algorithm difference, so CP-ANN model is identical as modeling spectral region used in DA model, i.e., the modeling light in table 4
3819~581cm of spectral limit-1It has been the modeling spectral region after optimization.
The main modeling parameters and estimated performance of 4 Chinese ephedra place of production CP-ANN model of table
4.2.3 the selection of the dimensionality reduction of spectroscopic data and modeling variable (principal component)
PCA dimensionality reduction is carried out to the spectroscopic data in selected modeling spectral region, when being modeled using different principal components, mould
The performance of type is there are notable difference, such as the model 1,11 and 12 in table 4, finally according to contribution rate sequential selection from high to low before
6 principal components predict the CP-ANN model in the Chinese ephedra place of production to establish.
4.2.4 the foundation and verifying of model
The CP-ANN with the verifying prediction Chinese ephedra place of production is established using the principal component scores of calibration set and verifying collection spectroscopic data
Model, (model 1 in table 4) as shown in Figure 4, calibration set right judging rate, cross validation right judging rate and verifying collection right judging rate are
100.0%, illustrate that the model has good estimated performance.
By Fig. 3 and 4 as it can be seen that linear DA model and non-linear CP-ANN model that the present embodiment is established can it is sensitive,
Accurately predict the place of production of Chinese ephedra, it was demonstrated that the highly sensitive spectrum of Chinese ephedra sample that the method for the present invention obtains contains the feature letter of sample
Breath.When Pretreated spectra Scheme Choice is to pre-process through NP, Pretreated spectra scheme that two kinds of models use, modeling spectrum model
Enclose all the same with Modelling feature variable, and two kinds of models all have preferable estimated performance, have confirmed the present invention each other from infrared
The effective information for accurately extracting and being utilized target signature in spectrum ensure that specificity, the sensitivity of analysis method of the present invention
And accuracy.
A kind of middle infrared transmission spectra method of the sensitive prediction Chinese ephedra of embodiment 3 in a few days plucking time
The present embodiment uses preparation method of test article same as Example 1, spectral measurement condition and method, with embodiment
1 difference is that the present embodiment uses the Chinese ephedra plant sample of in a few days different time picking to establish Chinese ephedra with preferred modeling parameters
The in a few days prediction model of plucking time.
1. the collection of Chinese ephedra plant sample
Collect and record the Shanxi ephedra sinica plant sample of in a few days different time picking, the Shanxi including the picking of 8 batches of mornings
The Shanxi ephedra sinica of ephedra sinica and picking in 7 batches of afternoons, placing under room temperature and 40%~60% relative humidities plants Chinese ephedra
Object sample contained humidity is reduced to metastable level, takes its herbaceous stem stem, obtains Chinese ephedra sample.
2. the preparation of Ephedra sample
With embodiment 1.
3. the preparation of test sample and the measurement of middle infrared spectrum
With embodiment 1.
4. the extraction and modeling of spectral signature variable
6 batches of mornings are chosen respectively from the Shanxi ephedra sinica that Shanxi ephedra sinica and 7 batches of afternoons that 8 batches of mornings pick are picked to adopt
The Shanxi ephedra sinica plucked and the Shanxi ephedra sinica of picking in 5 batches of afternoons are as calibration set sample, remaining is as verifying collection sample.
The selection and estimated performance of the main modeling parameters of 4.1DA model
4.1.1 the selection of Pretreated spectra scheme
Establish predict the Chinese ephedra in a few days DA model of plucking time when, in order to obtain the optimum prediction performance of model, to NP,
The multiple spectrums preconditioning technique such as MSC, SNV, FD, SD, SGS, NDS is screened and has been combined, and is shown in Table 5.The result shows that gained
When spectrum is pre-processed through MSC or SNV, the estimated performance of built DA model is optimal.
4.1.2 the selection of spectral region is modeled
It is automatic in modeling software to be modeled according to tested characteristic using aforementioned preferred Pretreated spectra scheme for spectral region
Artificial optimization is 3815 ± 117cm of wave number upper limit value on the basis of selection-1That is 3932~3698cm-1, wave number lower limit value 585 ±
117cm-1That is 702~468cm-1When, the calibration set right judging rate and verifying collection right judging rate of model can reach 100.0%, so excellent
Changing the spectral region in range can be used to model, such as the model 2,14 and 15 in table 5；And when modeling spectral region is not excellent
When changing in range, the calibration set right judging rate or/and verifying collection right judging rate of model are not up to 100.0%, such as 13 He of model in table 5
16。
The main modeling parameters and estimated performance of 5 Chinese ephedra of table in a few days plucking time DA model
4.1.3 the selection of the dimensionality reduction of spectroscopic data and modeling variable (principal component)
PCA dimensionality reduction is carried out to the spectroscopic data in selected modeling spectral region, when being modeled using different principal components, mould
The performance of type is there are notable difference, such as the model 2,11 and 12 in table 5, finally according to contribution rate sequential selection from high to low before
4 principal components establish the DA model of prediction Chinese ephedra in a few days plucking time.
4.1.4 the foundation and verifying of model
It is established using the principal component scores of calibration set and verifying collection spectroscopic data and verifies prediction Chinese ephedra in a few days plucking time
DA model, as shown in Figure 5 (model 2 in table 5), calibration set right judging rate and verifying collection right judging rate are 100.0%, explanation
The model has good estimated performance.
The selection and estimated performance of the main modeling parameters of 4.2CP-ANN model
4.2.1 the selection of Pretreated spectra scheme
When establishing the prediction Chinese ephedra in a few days CP-ANN model of plucking time, in order to obtain the optimum prediction performance of model,
The multiple spectrums preconditioning technique such as NP, MSC, SNV, FD, SD, SGS, NDS is screened and combined, is shown in Table 6.As a result table
Bright, when gained spectrum is pre-processed through MSC, the estimated performance of built CP-ANN model is optimal.
The main modeling parameters and estimated performance of 6 Chinese ephedra of table in a few days plucking time CP-ANN model
4.2.2 the selection of spectral region is modeled
It is identical as DA model as used sample spectra when verifying CP-ANN model due to establishing, and two kinds of models
Prediction characteristic is similarly the in a few days plucking time of Chinese ephedra, i.e., the effective information of same target feature is characterized in same spectra identical
Region, only modeling algorithm is different, so CP-ANN model is identical as modeling spectral region used in DA model, i.e., in table 6
3815~585cm of modeling spectral region-1It has been the modeling spectral region after optimization.
4.2.3 the selection of the dimensionality reduction of spectroscopic data and modeling variable (principal component)
PCA dimensionality reduction is carried out to the spectroscopic data in selected modeling spectral region, when being modeled using different principal components, mould
The performance of type is there are notable difference, such as the model 2,11 and 12 in table 6, finally according to contribution rate sequential selection from high to low before
4 principal components establish the CP-ANN model of prediction Chinese ephedra in a few days plucking time.
4.2.4 the foundation and verifying of model
It is established using the principal component scores of calibration set and verifying collection spectroscopic data and verifies prediction Chinese ephedra in a few days plucking time
CP-ANN model, as shown in Figure 6 (model 2 in table 6), calibration set right judging rate, cross validation right judging rate and verifying collection are just
Sentencing rate is 100.0%, illustrates that the model has good estimated performance.
By Figures 5 and 6 as it can be seen that linear DA model and non-linear CP-ANN model that the present embodiment is established can it is sensitive,
Accurately predict the in a few days plucking time of Chinese ephedra, it was demonstrated that the highly sensitive spectrum of Chinese ephedra sample that the method for the present invention obtains contains sample
Characteristic information.When Pretreated spectra Scheme Choice is MSC, Pretreated spectra scheme that two kinds of models use, modeling spectrum
Range and Modelling feature variable are all the same, and two kinds of models all have preferable estimated performance, have confirmed each other of the invention from red
The effective information for accurately extracting and being utilized target signature in external spectrum ensure that the specific, sensitive of analysis method of the present invention
Degree and accuracy.
A kind of middle infrared transmission spectra method of sensitive prediction Chinese ephedra Ephedrine content of embodiment 4
The present embodiment uses preparation method of test article same as Example 1, spectral measurement condition and method, with embodiment
1 difference is that the present embodiment uses ephedra sinica sample and its Content of Ephedrine With reference value to establish Chinese ephedra with preferred modeling parameters
The Quantitative Prediction Model of Ephedrine.
1. the collection of Chinese ephedra plant sample
25 batches of ephedra sinica plant samples are collected and record, placing under room temperature and 40%~60% relative humidities makes fiber crops
Yellow plant sample contained humidity is reduced to metastable level, takes its herbaceous stem stem, obtains Chinese ephedra sample.
2. the preparation of Ephedra sample and the measurement of Content of Ephedrine With reference value
The preparation method is the same as that of Example 1 for Ephedra sample.
Content reference value (the measurement of 25 batches of Ephedra sample Ephedrines is measured using high performance liquid chromatography (HPLC) method
Method bibliography: Lin Kai, Fan Qi, Yang Chenggang, Deng open English .RP-HPLC method measurement Chinese ephedra Ephedrine, pseudoephedrine and methyl
Ephedrine [J] Chinese herbal medicine .2006,37 (2): 282-284.), result be 0.68~15.39mg/g, i.e., 0.068%~
1.539%.
3. the preparation of test sample and the measurement of middle infrared spectrum
With embodiment 1.
4. the extraction and modeling of spectral signature variable
19 batches are chosen from 25 batches of ephedra sinica samples and is used as calibration set sample, remaining is as verifying collection sample.
The selection and estimated performance of the main modeling parameters of 4.1PLS model
4.1.1 the selection of Pretreated spectra scheme
It is right in order to obtain the optimum prediction performance of model when establishing the PLS model of prediction Chinese ephedra Ephedrine content
NP, MSC, SNV, FD, SD, SGS, NDS, MC, variance calibration are that the multiple spectrums preconditioning techniques such as VS are screened and combined,
It is shown in Table 7.The result shows that the estimated performance of built PLS model is optimal when gained spectrum is pre-processed through SD+MC.
The main modeling parameters and estimated performance of 7 Chinese ephedra Ephedrine content PLS model of table
4.1.2 the selection of spectral region is modeled
It is automatic in modeling software to be modeled according to tested characteristic using aforementioned preferred Pretreated spectra scheme for spectral region
Artificial optimization is 3986 ± 14cm of wave number upper limit value on the basis of selection-1That is 4000~3972cm-1, wave number lower limit value 414 ±
14cm-1That is 428~400cm-1When, the R of modelCWith RVIt is smaller that 0.8, RMSEC and RMSEV can be greater than, Bias 0mg/g,
So the spectral region in optimization range can be used to model, such as the model 12,18,19 and 20 in table 7；And when modeling spectrum
When range is not in optimization range, the R of modelCOr/and RVIt is larger less than 0.8, RMSEC or/and RMSEV, such as model 21 in table 7.
4.1.3 the selection of the dimensionality reduction of spectroscopic data and modeling variable (main gene)
PCA dimensionality reduction is carried out to the spectroscopic data in selected modeling spectral region, when being modeled using different main genes, mould
The performance of type is there are notable difference, such as the model 12 and 13 in table 7, finally according to before contribution rate sequential selection from high to low 4
A main gene establishes the PLS model of prediction Chinese ephedra Ephedrine content.
4.1.4 the foundation and verifying of model
Using calibration set with verifying collection spectroscopic data main gene score and corresponding Content of Ephedrine With reference value establish with
The PLS model of verifying prediction Chinese ephedra Ephedrine content, (model 12 in table 7) as shown in Figure 7, calibration set and verifying collection
Linear relationship is significant, root-mean-square error is smaller, and deviation is also smaller, illustrates that the model has good estimated performance.
The selection and estimated performance of the main modeling parameters of 4.2BP-ANN model
4.2.1 the selection of Pretreated spectra scheme
When establishing the BP-ANN model of prediction Chinese ephedra Ephedrine content, in order to obtain the optimum prediction performance of model,
The multiple spectrums preconditioning technique such as NP, MSC, SNV, FD, SD, SGS, NDS, MC, VS is screened and combined, is shown in Table 8.Knot
When fruit shows that gained spectrum is pre-processed through SD+MC, the estimated performance of built BP-ANN model is optimal.
4.2.2 the selection of spectral region is modeled
Used sample spectra is identical as PLS model when due to establishing and verifying BP-ANN model, and two kinds of models
Prediction characteristic be similarly the content of Chinese ephedra Ephedrine, i.e., the effective information of same target feature is characterized in same spectra in phase
Same region, only modeling algorithm is different, so BP-ANN model is identical as spectral region is modeled used in PLS model, i.e. table 8
In 3986~414cm of modeling spectral region-1It has been the modeling spectral region after optimization.
4.2.3 the selection of the dimensionality reduction of spectroscopic data and modeling variable (main gene)
PCA dimensionality reduction is carried out to the spectroscopic data in selected modeling spectral region, when being modeled using different main genes, mould
The performance of type is there are notable difference, such as the model 12 and 13 in table 8, finally according to before contribution rate sequential selection from high to low 4
A main gene establishes the BP-ANN model of prediction Chinese ephedra Ephedrine content.
4.2.4 the foundation and verifying of model
Using calibration set with verifying collection spectroscopic data main gene score and corresponding Content of Ephedrine With reference value establish with
The BP-ANN model of verifying prediction Chinese ephedra Ephedrine content, (model 12 in table 8) as shown in Figure 8, calibration set and verifying
The linear relationship of collection is significant, root-mean-square error is smaller, illustrates that the model has good estimated performance.
The main modeling parameters and estimated performance of 8 Chinese ephedra Ephedrine content BP-ANN model of table
By Fig. 7 and 8 as it can be seen that Linear PLS model and non-linear BP-ANN model that the present embodiment is established can it is sensitive,
Accurately predict the content of Chinese ephedra Ephedrine, it was demonstrated that the highly sensitive spectrum of Chinese ephedra sample that the method for the present invention obtains contains sample
Characteristic information.Pretreated spectra scheme, modeling spectral region and the Modelling feature variable used due to two kinds of models is all the same,
And two kinds of models all have preferable estimated performance, have confirmed the present invention each other and have accurately extracted and be utilized mesh from infrared spectroscopy
The effective information for marking feature, ensure that specificity, sensitivity and the accuracy of analysis method of the present invention.
The middle infrared transmission spectra method of Determination of Pseudoephedrine in a kind of sensitive prediction Chinese ephedra of embodiment 5
The present embodiment uses sample same as Example 4, preparation method of test article, spectral measurement condition and method, with
The difference of embodiment 4 is that the present embodiment uses Determination of Pseudoephedrine reference value in ephedra sinica sample to build with preferred modeling parameters
The Quantitative Prediction Model of pseudoephedrine in vertical Chinese ephedra.
1. the collection of Chinese ephedra plant sample
With embodiment 4.
2. the preparation of Ephedra sample and the measurement of Determination of Pseudoephedrine reference value
The preparation method is the same as that of Example 1 for Ephedra sample.
Using content reference value (the measuring method bibliography: woods of pseudoephedrine in HPLC method measurement Ephedra sample
Triumphant, Fan Qi, Yang Chenggang, Deng open English .RP-HPLC method measurement Chinese ephedra Ephedrine, pseudoephedrine and methylephedrine [J] medium-height grass
Medicine .2006,37 (2): 282-284.), result be 0.23~5.07mg/g, i.e., 0.023%~0.507%.
3. the preparation of test sample and the measurement of middle infrared spectrum
With embodiment 1.
4. the extraction and modeling of spectral signature variable
19 batches are chosen from 25 batches of ephedra sinica samples and is used as calibration set sample, remaining is as verifying collection sample.
The selection and estimated performance of the main modeling parameters of 4.1PLS model
4.1.1 the selection of Pretreated spectra scheme
It is right in order to obtain the optimum prediction performance of model in establishing prediction Chinese ephedra when the PLS model of Determination of Pseudoephedrine
The multiple spectrums preconditioning technique such as NP, MSC, SNV, FD, SD, SGS, NDS, MC, VS is screened and has been combined, and is shown in Table 9.As a result
When showing that gained spectrum is pre-processed through SD+MC, the estimated performance of built PLS model is optimal.
4.1.2 the selection of spectral region is modeled
It is automatic in modeling software to be modeled according to tested characteristic using aforementioned preferred Pretreated spectra scheme for spectral region
Artificial optimization is 3930 ± 70cm of wave number upper limit value on the basis of selection-1That is 4000~3860cm-1, wave number lower limit value 470 ±
70cm-1That is 540~400cm-1When, the R of modelCWith RVIt is smaller that 0.8, RMSEC and RMSEV can be greater than, Bias 0mg/g,
So the spectral region in optimization range can be used to model, such as the model 12,18,19 and 20 in table 9；And when modeling spectrum
When range is not in optimization range, the R of modelCOr/and RVIt is larger less than 0.8, RMSEC or/and RMSEV, such as the model in table 9
21。
The main modeling parameters and estimated performance of Determination of Pseudoephedrine PLS model in 9 Chinese ephedra of table
4.1.3 the selection of the dimensionality reduction of spectroscopic data and modeling variable (main gene)
PCA dimensionality reduction is carried out to the spectroscopic data in selected modeling spectral region, when being modeled using different main genes, mould
The performance of type is there are notable difference, such as the model 12 and 13 in table 9, finally according to before contribution rate sequential selection from high to low 4
A main gene establishes the PLS model of Determination of Pseudoephedrine in prediction Chinese ephedra.
4.1.4 the foundation and verifying of model
It is established using calibration set with the main gene score of verifying collection spectroscopic data and corresponding Determination of Pseudoephedrine reference value
With the PLS model of Determination of Pseudoephedrine in verifying prediction Chinese ephedra, (9 model 12 of table) as shown in Figure 9, calibration set and verifying collection
Linear relationship is significant, root-mean-square error is smaller, and deviation is also smaller, illustrates that the model has good estimated performance.
The selection and estimated performance of the main modeling parameters of 4.2BP-ANN model
4.2.1 the selection of Pretreated spectra scheme
In establishing prediction Chinese ephedra when the BP-ANN model of Determination of Pseudoephedrine, in order to obtain the optimum prediction of model
Can, the multiple spectrums preconditioning technique such as NP, MSC, SNV, FD, SD, SGS, NDS, MC, VS is screened and combined, is shown in Table
10.The result shows that the estimated performance of built BP-ANN model is optimal when gained spectrum is pre-processed through SD+MC.
The main modeling parameters and estimated performance of Determination of Pseudoephedrine BP-ANN model in 10 Chinese ephedra of table
4.2.2 the selection of spectral region is modeled
Used sample spectra is identical as PLS model when due to establishing and verifying BP-ANN model, and two kinds of models
Prediction characteristic be similarly the content of pseudoephedrine in Chinese ephedra, i.e., the effective information that same target feature is characterized in same spectra exists
Same area, only modeling algorithm is different, so BP-ANN model is identical as modeling spectral region used in PLS model, i.e.,
3930~470cm of modeling spectral region in table 10-1It has been the modeling spectral region after optimization.
4.2.3 the selection of the dimensionality reduction of spectroscopic data and modeling variable (main gene)
PCA dimensionality reduction is carried out to the spectroscopic data in selected modeling spectral region, when being modeled using different main genes, mould
The performance of type is there are notable difference, such as the model 12 and 13 in table 10, finally according to before contribution rate sequential selection from high to low 4
A main gene establishes the BP-ANN model of Determination of Pseudoephedrine in prediction Chinese ephedra.
4.2.4 the foundation and verifying of model
It is established using calibration set with the main gene score of verifying collection spectroscopic data and corresponding Determination of Pseudoephedrine reference value
It is as shown in Figure 10 (10 model 12 of table) with the BP-ANN model of Determination of Pseudoephedrine in verifying prediction Chinese ephedra, calibration set with test
The linear relationship of card collection is significant, root-mean-square error is smaller, illustrates that the model has good estimated performance.
By Fig. 9 and 10 as it can be seen that Linear PLS model and non-linear BP-ANN model that the present embodiment is established being capable of spirits
It is quick, accurately predict Chinese ephedra in pseudoephedrine content, it was demonstrated that the method for the present invention obtain the highly sensitive spectrum of Chinese ephedra sample include
The characteristic information of sample.Pretreated spectra scheme, modeling spectral region and the Modelling feature variable used due to two kinds of models
It is all the same, and two kinds of models all have preferable estimated performance, confirmed each other the present invention accurately extracted from infrared spectroscopy and
The effective information of target signature is utilized, ensure that specificity, sensitivity and the accuracy of analysis method of the present invention.
The middle infrared transmission spectra method of methylephedrine content in a kind of sensitive prediction Chinese ephedra of embodiment 6
The present embodiment uses sample same as Example 4, preparation method of test article, spectral measurement condition and method, with
The difference of embodiment 4 is that the present embodiment uses in ephedra sinica sample methylephedrine content reference value with preferred modeling parameters
Establish the Quantitative Prediction Model of methylephedrine in Chinese ephedra.
1. the collection of Chinese ephedra plant sample
With embodiment 4.
2. the preparation of Ephedra sample and the measurement of methylephedrine content reference value
The preparation method is the same as that of Example 1 for Ephedra sample.
Using HPLC method measurement Ephedra sample in methylephedrine content reference value (measuring method bibliography:
Lin Kai, Fan Qi, Yang Chenggang, Deng open in English .RP-HPLC method measurement Chinese ephedra Ephedrine, pseudoephedrine and methylephedrine [J]
Herbal medicine .2006,37 (2): 282-284.), result be 0.082~0.966mg/g, i.e., 0.0082%~0.0966%.
3. the preparation of test sample and the measurement of middle infrared spectrum
With embodiment 1.
4. the extraction and modeling of spectral signature variable
19 batches are chosen from 25 batches of ephedra sinica samples and is used as calibration set sample, remaining is as verifying collection sample.
The selection and estimated performance of the main modeling parameters of 4.1PLS model
4.1.1 the selection of Pretreated spectra scheme
In establishing prediction Chinese ephedra when the PLS model of methylephedrine content, in order to obtain the optimum prediction performance of model,
The multiple spectrums preconditioning technique such as NP, MSC, SNV, FD, SD, SGS, NDS, MC, VS is screened and combined, is shown in Table 11.
The result shows that the estimated performance of built PLS model is optimal when gained spectrum is pre-processed through FD+MC.
4.1.2 the selection of spectral region is modeled
It is automatic in modeling software to be modeled according to tested characteristic using aforementioned preferred Pretreated spectra scheme for spectral region
Artificial optimization is 3800 ± 200cm of wave number upper limit value on the basis of selection-1That is 4000~3600cm-1, wave number lower limit value 600 ±
200cm-1That is 800~400cm-1When, the R of modelCWith RVIt is smaller that 0.8, RMSEC and RMSEV can be greater than, Bias 0mg/g,
So the spectral region in optimization range can be used to model, such as the model 11,18,19 and 20 in table 11；And when modeling light
When spectral limit is not in optimization range, the R of modelCOr/and RVIt is larger less than 0.8, RMSEC or/and RMSEV, such as the mould in table 11
Type 21.
4.1.3 the selection of the dimensionality reduction of spectroscopic data and modeling variable (main gene)
PCA dimensionality reduction is carried out to the spectroscopic data in selected modeling spectral region, when being modeled using different main genes, mould
The performance of type is there are notable difference, such as the model 11 and 12 in table 11, finally according to before contribution rate sequential selection from high to low 4
A main gene establishes the PLS model of methylephedrine content in prediction Chinese ephedra.
4.1.4 the foundation and verifying of model
It is built using calibration set with the main gene score of verifying collection spectroscopic data and corresponding methylephedrine content reference value
The vertical PLS model with methylephedrine content in verifying prediction Chinese ephedra, (model 11 in table 11) as shown in figure 11, calibration set
The linear relationship collected with verifying is significant, root-mean-square error is smaller, and deviation is also smaller, illustrates that the model has good predictability
Energy.
The main modeling parameters and estimated performance of methylephedrine content PLS model in 11 Chinese ephedra of table
The selection and estimated performance of the main modeling parameters of 4.2BP-ANN model
4.2.1 the selection of Pretreated spectra scheme
In establishing prediction Chinese ephedra when the BP-ANN model of methylephedrine content, in order to obtain the optimum prediction of model
Can, the multiple spectrums preconditioning technique such as NP, MSC, SNV, FD, SD, SGS, NDS, MC, VS is screened and combined, is shown in Table
12.The result shows that the estimated performance of built BP-ANN model is optimal when gained spectrum is pre-processed through FD+MC.
The main modeling parameters and estimated performance of methylephedrine content BP-ANN model in 12 Chinese ephedra of table
4.2.2 the selection of spectral region is modeled
Used sample spectra is identical as PLS model when due to establishing and verifying BP-ANN model, and two kinds of models
Prediction characteristic be similarly the content of methylephedrine in Chinese ephedra, i.e., the effective information of same target feature is characterized in same spectra
In same area, only modeling algorithm is different, so BP-ANN model is identical as modeling spectral region used in PLS model,
3800~600cm of modeling spectral region i.e. in table 12-1It has been the modeling spectral region after optimization.
4.2.3 the selection of the dimensionality reduction of spectroscopic data and modeling variable (main gene)
PCA dimensionality reduction is carried out to the spectroscopic data in selected modeling spectral region, when being modeled using different main genes, mould
The performance of type is there are notable difference, such as the model 11 and 12 in table 12, finally according to before contribution rate sequential selection from high to low 4
A main gene establishes the BP-ANN model of methylephedrine content in prediction Chinese ephedra.
4.2.4 the foundation and verifying of model
It is built using calibration set with the main gene score of verifying collection spectroscopic data and corresponding methylephedrine content reference value
The vertical BP-ANN model with methylephedrine content in verifying prediction Chinese ephedra, (model 11 in table 12) as shown in figure 12, school
The linear relationship of positive collection and verifying collection is significant, root-mean-square error is smaller, illustrates the model with good estimated performance.
By Figure 11 and 12 as it can be seen that Linear PLS model and non-linear BP-ANN model that the present embodiment is established being capable of spirits
Content that is quick, accurately predicting methylephedrine in Chinese ephedra, it was demonstrated that the highly sensitive spectrum packet of Chinese ephedra sample that the method for the present invention obtains
The characteristic information of sample is contained.Pretreated spectra scheme, modeling spectral region and the Modelling feature used due to two kinds of models is become
Measure all the same, and two kinds of models all have preferable estimated performance, have confirmed the present invention each other and have accurately extracted from infrared spectroscopy
With the effective information that target signature is utilized, specificity, sensitivity and the accuracy of analysis method of the present invention ensure that.
By above-mentioned experimental result it is found that institute's construction method of the present invention can not only sensitive, accurately predict same morning day or
The small Chinese ephedra quality characteristic of difference of picking in afternoon also can sensitive, accurately predict in Chinese ephedra content down to 0.0082%
Methylephedrine content.Therefore, the method for the present invention can individually or simultaneously sensitive, accurately predict that a variety of qualities of Chinese ephedra are special
Property, the secondary metabolite including identifying small difference and quantitative forecast low content.
Finally, it is noted that the above examples are only used to illustrate the technical scheme of the present invention and are not limiting, although passing through
Invention has been described referring to the preferred embodiment of the present invention, however, those skilled in the art should understand that, can be
Various changes are made to it in form and in details, without departing from guarantor of the invention defined by the appended claims
Protect range.
Claims (4)
1. a kind of mid-infrared light spectrometry of sensitive prediction Chinese ephedra quality characteristic, which is characterized in that the quality characteristic is type, produces
Ground, in a few days plucking time and secondary metabolites content are any one or more of；Wherein,
Predict Chinese ephedra type mid-infrared light spectrometry the following steps are included:
(1) it collects and records different types of Chinese ephedra plant sample, it is dry, its herbaceous stem stem is taken, Chinese ephedra sample is obtained；The Chinese ephedra is planted
The type of object is ephedra sinica (Ephedra sinica Stapf), epheday intermedia (Ephedra intermedia Schrenk et
) or ephedra equisetina (Ephedra equisetina Bge.) C.A.Mey.；
(2) by Chinese ephedra sample comminution, sieving obtains Ephedra sample；
(3) Ephedra sample is mixed with potassium bromide, suppresses, obtains Ephedra kbr tablet, as test sample, use infrared light
Spectrometer is in 4000~400cm-1The middle infrared transmission spectra of measurement test sample in range, every time with identical parameters before scanning test sample
Scan simultaneously background correction；
(4) gained spectrum is not preprocessed or is smoothly SGS pretreatment through Savitzky-Golay, chooses modeling spectral region
Wave number upper limit value is 3936 ± 64cm-1That is 4000~3872cm-1, wave number lower limit value be 464 ± 64cm-1That is 528~400cm-1,
Using Principal Component Analysis, that is, PCA dimensionality reduction, preceding 8 principal components are chosen according to the sequence of contribution rate from high to low, establish prediction fiber crops
The discriminant analysis of yellow race's class, that is, DA model；
Alternatively, gained spectrum chooses identical modeling spectral region and identical principal component after first derivative, that is, FD pretreatment,
It establishes the opposite of prediction Chinese ephedra type and propagates artificial neural network, that is, CP-ANN model；
(5) the Chinese ephedra sample for taking unknown type, by step (2), (3) and (4) the method prepare test sample, in measurement it is infrared
It penetrates spectrum and carries out spectroscopic data processing, then applying step (4) model built predicts the type of unknown Chinese ephedra sample；
Predict the Chinese ephedra place of production mid-infrared light spectrometry the following steps are included:
(1) the same type Chinese ephedra plant sample of different sources is collected and records, it is dry, its herbaceous stem stem is taken, Chinese ephedra sample is obtained；It is described
The place of production of Chinese ephedra plant is Shanxi or the Inner Mongol；
(2) by Chinese ephedra sample comminution, sieving obtains Ephedra sample；
(3) Ephedra sample is mixed with potassium bromide, suppresses, obtains Ephedra kbr tablet, as test sample, use infrared light
Spectrometer is in 4000~400cm-1The middle infrared transmission spectra of measurement test sample in range, every time with identical parameters before scanning test sample
Scan simultaneously background correction；
(4) gained spectrum is not preprocessed or pre-process through SGS, and the wave number upper limit value for choosing modeling spectral region is 3819 ±
181cm-1That is 4000~3638cm-1, wave number lower limit value be 581 ± 181cm-1That is 762~400cm-1, using PCA dimensionality reduction, according to
The sequence of contribution rate from high to low chooses preceding 6 principal components, establishes the DA model in the prediction Chinese ephedra place of production；Alternatively, gained spectrum is not
It is preprocessed, identical modeling spectral region and identical principal component are chosen, the CP-ANN model in the prediction Chinese ephedra place of production is established；
(5) the Chinese ephedra sample for taking the unknown place of production, by step (2), (3) and (4) the method prepare test sample, in measurement it is infrared
It penetrates spectrum and carries out spectroscopic data processing, then applying step (4) model built predicts the place of production of unknown Chinese ephedra sample；
Predict Chinese ephedra in a few days plucking time mid-infrared light spectrometry the following steps are included:
(1) the similar place of production Chinese ephedra plant sample of the same race of different in a few days plucking times is collected and records, it is dry, its herbaceous stem stem is taken,
Obtain Chinese ephedra sample；The plucking time of the Chinese ephedra plant is the morning or afternoon of same day；
(2) by Chinese ephedra sample comminution, sieving obtains Ephedra sample；
(3) Ephedra sample is mixed with potassium bromide, suppresses, obtains Ephedra kbr tablet, as test sample, use infrared light
Spectrometer is in 4000~400cm-1The middle infrared transmission spectra of measurement test sample in range, every time with identical parameters before scanning test sample
Scan simultaneously background correction；
(4) gained spectrum is SNV pretreatment through multiplicative scatter correction, that is, MSC or standard normal transformation, chooses modeling spectral region
Wave number upper limit value be 3815+117cm-1That is 3932~3698cm-1, wave number lower limit value be 585 ± 117cm-1I.e. 702~
468cm-1, using PCA dimensionality reduction, preceding 4 principal components are chosen according to the sequence of contribution rate from high to low, prediction Chinese ephedra is established and in a few days adopts
Pluck the DA model of time；
Alternatively, gained spectrum is pre-processed through MSC, identical modeling spectral region and identical principal component are chosen, establishes prediction fiber crops
The CP-ANN model of yellow in a few days plucking time；
(5) the Chinese ephedra sample for taking plucking time in unknown day prepares test sample, measurement by step (2), (3) and (4) the method
Middle infrared transmission spectra simultaneously carries out spectroscopic data processing, and then applying step (4) model built predicts the day of unknown Chinese ephedra sample
Interior plucking time；
Predict Chinese ephedra in secondary metabolites content mid-infrared light spectrometry the following steps are included:
(1) the same type Chinese ephedra plant sample of different secondary metabolites contents is collected and records, it is dry, its herbaceous stem stem is taken, is obtained
Chinese ephedra sample；The secondary metabolite is that ephedrine, d-pseudo-ephedrine and methylephedrine are any one or more of；
(2) by Chinese ephedra sample comminution, sieving obtains Ephedra sample；Measure secondary metabolites content in Ephedra sample
Reference value；
(3) Ephedra sample is mixed with potassium bromide, suppresses, obtains Ephedra kbr tablet, as test sample, use infrared light
Spectrometer is in 4000~400cm-1The middle infrared transmission spectra of measurement test sample in range, every time with identical parameters before scanning test sample
Scan simultaneously background correction；
(4) when establishing the prediction model of Chinese ephedra Ephedrine content, gained spectrum is through second dervative, that is, SD and mean value centralization
After MC pretreatment, the wave number upper limit value for choosing modeling spectral region is 3986 ± 14cm-1That is 4000~3972cm-1, wave number lower limit
Value is 414 ± 14cm-1That is 428~400cm-1, using PCA dimensionality reduction, preceding 4 masters are chosen according to the sequence of contribution rate from high to low
The factor establishes the Partial Least Squares Regression, that is, PLS or back-propagation artificial neural network, that is, BP- of prediction Chinese ephedra Ephedrine content
ANN model；
When establishing the prediction model of Determination of Pseudoephedrine in Chinese ephedra, gained spectrum is pre-processed through SD and MC, chooses modeling spectrum model
The wave number upper limit value enclosed is 3930 ± 70cm-1That is 4000~3860cm-1, wave number lower limit value be 470 ± 70cm-1I.e. 540~
400cm-1, using PCA dimensionality reduction, preceding 4 main genes are chosen according to the sequence of contribution rate from high to low, establish pseudo- fiber crops in prediction Chinese ephedra
PLS the or BP-ANN model of yellow alkali content；
When establishing the prediction model of methylephedrine content in Chinese ephedra, FD and MC is carried out to gained spectrum and is pre-processed, modeling is chosen
The wave number upper limit value of spectral region is 3800 ± 200cm-1That is 4000~3600cm-1, wave number lower limit value be 600 ± 200cm-1I.e.
800~400cm-1, using PCA dimensionality reductions, preceding 4 main genes are chosen according to the sequence of contribution rate from high to low, establish prediction Chinese ephedra
PLS the or BP-ANN model of middle methylephedrine content；
(5) the Chinese ephedra sample for taking unknown secondary metabolites content, by step (2), (3) and (4) the method prepare test sample,
Infrared transmission spectra and spectroscopic data processing is carried out in measurement, then applying step (4) model built predicts unknown Chinese ephedra sample
Secondary metabolites content.
2. the mid-infrared light spectrometry of the sensitive prediction Chinese ephedra quality characteristic of one kind according to claim 1, it is characterised in that: institute
State prediction Chinese ephedra type, the place of production, in a few days plucking time or/and secondary metabolites content mid-infrared light spectrometry the step of (2)
In, crossing grit number is 100~200 mesh.
3. the mid-infrared light spectrometry of the sensitive prediction Chinese ephedra quality characteristic of one kind according to claim 1 or 2, feature exist
In: it is described prediction Chinese ephedra type, the place of production, in a few days plucking time or/and secondary metabolites content mid-infrared light spectrometry step
It suddenly is in mass ratio to mix at 1: 150 with potassium bromide by Ephedra sample in (3), tableting pressure is 10~30MPa, when tabletting
Between be 1~2 minute, Ephedra kbr tablet is made.
4. the mid-infrared light spectrometry of the sensitive prediction Chinese ephedra quality characteristic of one kind according to claim 1 or 2, feature exist
In: it is described prediction Chinese ephedra type, the place of production, in a few days plucking time or/and secondary metabolites content mid-infrared light spectrometry step
Suddenly in (3), the resolution ratio of spectral measurement is 4cm-1, scanning times are not less than 32 times.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810913774.1A CN109001143B (en) | 2018-08-01 | 2018-08-01 | Intermediate infrared spectroscopy for sensitively predicting quality characteristics of ephedra |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810913774.1A CN109001143B (en) | 2018-08-01 | 2018-08-01 | Intermediate infrared spectroscopy for sensitively predicting quality characteristics of ephedra |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109001143A true CN109001143A (en) | 2018-12-14 |
| CN109001143B CN109001143B (en) | 2022-02-11 |
Family
ID=64594824
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810913774.1A Active CN109001143B (en) | 2018-08-01 | 2018-08-01 | Intermediate infrared spectroscopy for sensitively predicting quality characteristics of ephedra |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109001143B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110220863A (en) * | 2019-06-25 | 2019-09-10 | 湖南中医药大学 | A kind of discrimination method of honeysuckle and Honeysuckle flower Chinese materia medica preparation based on ATR-FTIR |
| CN110346323A (en) * | 2019-07-30 | 2019-10-18 | 江西中医药大学 | A method of based on near-infrared spectrum technique on-line checking canopy powder concentrate |
| CN112861411A (en) * | 2019-11-27 | 2021-05-28 | 国能生物发电集团有限公司 | Biomass sodium content measurement and modeling method based on near infrared spectrum principal component and neural network |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101342199A (en) * | 2007-07-13 | 2009-01-14 | 未名天人中药有限公司 | Chinese ephedra prescription particle, preparation method and quality control method thereof |
| CN101435771A (en) * | 2008-07-15 | 2009-05-20 | 中国科学院上海应用物理研究所 | Efedrina for identifying different optical rotation performances by THz-TDS |
| CN105334183A (en) * | 2014-08-07 | 2016-02-17 | 重庆医科大学 | Method for identifying certifiable Herba Ephedrae based on near infrared spectroscopy |
| CN106353275A (en) * | 2015-07-24 | 2017-01-25 | 重庆医科大学 | Method for quickly measuring alkaloids of ephedra herb medicinal material based on ultraviolet spectrum |
| CN107917896A (en) * | 2017-11-29 | 2018-04-17 | 宁夏医科大学 | Radix glycyrrhizae method for quick identification based near infrared spectrum and Clustering Analysis Technology |
-
2018
- 2018-08-01 CN CN201810913774.1A patent/CN109001143B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101342199A (en) * | 2007-07-13 | 2009-01-14 | 未名天人中药有限公司 | Chinese ephedra prescription particle, preparation method and quality control method thereof |
| CN101435771A (en) * | 2008-07-15 | 2009-05-20 | 中国科学院上海应用物理研究所 | Efedrina for identifying different optical rotation performances by THz-TDS |
| CN105334183A (en) * | 2014-08-07 | 2016-02-17 | 重庆医科大学 | Method for identifying certifiable Herba Ephedrae based on near infrared spectroscopy |
| CN106353275A (en) * | 2015-07-24 | 2017-01-25 | 重庆医科大学 | Method for quickly measuring alkaloids of ephedra herb medicinal material based on ultraviolet spectrum |
| CN107917896A (en) * | 2017-11-29 | 2018-04-17 | 宁夏医科大学 | Radix glycyrrhizae method for quick identification based near infrared spectrum and Clustering Analysis Technology |
Non-Patent Citations (3)
| Title |
|---|
| 易珍奎等: "近红外漫反射光谱法结合CP-ANN和PLS高通量分析草麻黄药材", 《药物分析杂质》 * |
| 范琦: "傅里叶变换近红外漫反射光谱法结合多元分析技术判别麻黄植物的方法学研究", 《中国博士学位论文全文数据库 医药卫生科技辑》 * |
| 陈兰珍等: "《蜂蜜近红外光谱检测技术》", 31 January 2012, 中国轻工业出版社 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110220863A (en) * | 2019-06-25 | 2019-09-10 | 湖南中医药大学 | A kind of discrimination method of honeysuckle and Honeysuckle flower Chinese materia medica preparation based on ATR-FTIR |
| CN110346323A (en) * | 2019-07-30 | 2019-10-18 | 江西中医药大学 | A method of based on near-infrared spectrum technique on-line checking canopy powder concentrate |
| CN112861411A (en) * | 2019-11-27 | 2021-05-28 | 国能生物发电集团有限公司 | Biomass sodium content measurement and modeling method based on near infrared spectrum principal component and neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109001143B (en) | 2022-02-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101961360B (en) | Near infrared spectrum identification method for pseudo-ginseng | |
| CN108519348A (en) | Licorice medicinal materials Near-Infrared Quantitative Analysis model and detection method and standard | |
| CN103091274B (en) | The method of near-infrared diffuse-reflectance spectroscopic assay Radix Salviae Miltiorrhizae for injection Polyphenol Acids moisture | |
| CN104568822B (en) | A kind of capsule of weeping forsythia medicinal material multi objective while quick determination method | |
| CN109001143A (en) | A kind of mid-infrared light spectrometry of sensitive prediction Chinese ephedra quality characteristic | |
| CN109297929B (en) | Method for establishing quality grading of salvia miltiorrhiza decoction pieces by utilizing near infrared technology | |
| CN103969211B (en) | A kind of method using near infrared spectrum detection FUFANG DANSHEN PIAN moisture | |
| CN103033486B (en) | Method for near infrared spectrum monitoring of quality of pericarpium citri reticulatae and citrus chachiensis hortorum medicinal materials | |
| CN104062258B (en) | Method for rapid determination of soluble solids in compound ass-hide glue pulp by near infrared spectroscopy | |
| CN106841083A (en) | Sesame oil quality detecting method based on near-infrared spectrum technique | |
| CN107091815A (en) | A kind of method for identifying rosewood | |
| CN101780141A (en) | Method for measuring content of tannin in red sage root extracting solution by near infrared spectrum | |
| CN113008805B (en) | Radix angelicae decoction piece quality prediction method based on hyperspectral imaging depth analysis | |
| CN101961379A (en) | Near infrared spectrum identification method for red sage roots | |
| CN101231274A (en) | Method for rapid measuring allantoin content in yam using near infrared spectrum | |
| CN102313712B (en) | Correction method of difference between near-infrared spectrums with different light-splitting modes based on fiber material | |
| CN107917897A (en) | The method of the special doctor's food multicomponent content of near infrared ray | |
| CN106198446A (en) | The method of L-Borneol content near infrared spectrum quick test Herba Blumeae Balsamiferae leaf powder | |
| CN108663337A (en) | A kind of method and its application measuring tanshinone component | |
| CN104596980A (en) | Method for measuring hot water solvends of reconstituted tobacco by paper-making process by virtue of near infrared reflectance spectroscopy technique | |
| CN103760130A (en) | Method for measuring tween-80 content in compound musk injection by using near-infrared spectrum | |
| CN109540837A (en) | The method that near-infrared quickly detects Boehmeria nivea leaves wood fibre cellulose content | |
| CN108645809A (en) | A kind of method that near-infrared spectrum technique quickly identifies rosin original tree species | |
| WO2023207453A1 (en) | Traditional chinese medicine ingredient analysis method and system based on spectral clustering | |
| CN105334183A (en) | Method for identifying certifiable Herba Ephedrae based on near 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |