CN101299022A - Method for evaluating Chinese medicine comprehensive quality using near infrared spectra technique - Google Patents

Abstract

The present invention relates to a method for evaluating the comprehensive quality of material in Chinese medicine with near infrared spectrum technique. The fast detection to the comprehensive quality of material of Chinese medicine can be realized. The technical scheme for settling is firstly establishing a correction model, and the method comprises the following steps: collecting the sample of material of Chinese medicine as a correction collection sample, collecting the near infrared diffuse reflection spectrum of the sample, preprocessing the obtained spectrum, measuring the content of the index component with a detection method corresponding with the sample in the conventional method, combining the spectrum with the content of index component, applying a partial least square in the chemometrics for establishing a quantitative analysis model, scanning the near infrared spectrogram when the detected sample is crushed, and inputting the spectrogram into the quantitative analysis model. Then the content of index component in the material of Chinese medicine can be detected. The method according to the invention has the advantages of short required time in the whole process, high speed, accuracy, on-line measurement, and increased production efficiency. The man power and physical power can be greatly saved. Tremendous economic benefit and social benefit can be generated.

Description

Utilize near-infrared spectrum technique to estimate the method for Chinese medicine medicinal material overall quality

One, technical field

The present invention relates to field of medicaments, particularly a kind of method of utilizing near-infrared spectrum technique to estimate Chinese medicine medicinal material overall quality.

Two, background technology

Chinese medicine is the rarity of our Chinese nation, but because it is various in style, medicinal material separate sources, the place of production, growth year, collecting season, process of preparing Chinese medicine job operation are especially artificially manufactured the fake, are mixed chemical medicine etc. and all affect the traditional Chinese medicine quality quality, make the crude drug that circulates on the market that the true and false, good and bad difference be arranged; Conventional quality evaluating method sample preparation process complexity, checkout procedure is loaded down with trivial details, cost is high, efficient is low, the needs that can not adapt to the herbal pharmaceutical modernization development, and often control the quality of Chinese medicine, can not embody complicacy and its whole inherent quality of traditional Chinese medicine ingredients with single index components; And existing detection means mostly is offline inspection, and the shortcoming of hysteresis is arranged aborning.These all cause China's herbal pharmaceutical industry technology level not high, and production run lacks feasible method of quality control.

Three, summary of the invention

At above-mentioned situation, the present invention's purpose just provides a kind of method of utilizing near-infrared spectrum technique to estimate Chinese medicine medicinal material overall quality, can realize the fast detecting problem of Chinese crude drug overall quality, the technical scheme of its solution is, at first set up calibration model, method is to collect the Chinese crude drug sample as the calibration set sample, gather near infrared (NIR) diffuse reflection spectrum of sample, and gained spectrum carried out pre-service, then with recording its index component content with the corresponding detection method of sample in the conventional method, collection of illustrative plates is combined with the index component content, partial least square method in the Applied Chemometrics (PLS method) is set up Quantitative Analysis Model, and testing sample is pulverized its near infrared light spectrogram of back scanning, and collection of illustrative plates is imported Quantitative Analysis Model, can record the content of index composition in this Chinese crude drug, whole process required time is short, speed is fast, accurately, but on-line measurement is enhanced productivity, save the man power and material greatly, can produce tremendous economic and social benefit.

Four, embodiment

Below the specific embodiment of the present invention is elaborated.

The present invention realizes that by following steps at first set up calibration model, method is:

1) collects the calibration set sample: collect the different places of production, different cultivars, different collecting season, different concocting methods etc. and contain the medicinal material of each species diversity as the calibration set sample, also can be in the reality according to the scope of application of calibration model, its quantity can constantly increase, the traditional Chinese medicinal material samples crushing screening of collecting, standby;

2) gather sample spectrum: use ft-nir spectrometer, the near-infrared diffuse reflection spectrum of acquisition correction collection sample, test mode: integrating sphere diffuse reflection, scanning times 32～128, resolution 4～16cm ^-1, gain 1～8, the absorbance data form is: log1/R, spectral scan scope 12000～4000cm ^-1, obtain spectroscopic data, and sample be divided into calibration set and forecast set two classes;

3) sample spectrum pre-service: the spectroscopic data that scanning is obtained carries out pre-service, NIR spectrum is through single order or second-order differential, the filtering of Norris derivative, Savitzky-Golay is level and smooth, pre-service such as polynary scatter correction or standard canonical transformation, and with correlation spectrometry spectrogram is carried out wavelength and select, these preprocess methods can use separately, also can a plurality ofly use simultaneously, to reach best pretreating effect, the error that produces in the NIR spectral analysis is mainly from the high frequency random noise, baseline drift, at the bottom of the code book, sample is inhomogeneous, light scattering etc., for solving the interference that various factors produces spectrum, fully from spectrum, extract validity feature information, must carry out pre-service to spectrum, preprocess method is as follows:

1. many first scatter corrections: in near-infrared spectrum analysis, the difference of physical propertys such as the solid particle degree of measured object, crystalline form, the difference that can cause spectrogram, this species diversity enter solid interior the near infrared light process light path be absorbed the different of degree and cause, eliminating this scattering effect two kinds of methods commonly used is polynary scatter correction MSC;

2. derivative method: be preprocess method commonly used, commonly used to first order derivative and second derivative arranged, first order derivative can be eliminated the skew of spectrum baseline significantly, this is very effective to the uncertainty of proofreading and correct spectrum, adopt second derivative then can eliminate the linear tilt (also claiming the rotation of collection of illustrative plates) of baseline, after handling through second derivative as the broad peak in the former spectrum, can become very sharp-pointed, help in the peak shape of complexity, determining better the accurate position at peak like this, improved signal to noise ratio (S/N ratio);

In near infrared spectrum is region-wide, the spectral absorption information at different wave length place is different for the contributed value of setting up model at last, at some wavelength place, impurity absorbs and disturbs and is better than the absorption that target components produces greatly, and be difficult to characteristic information effectively be extracted by existing information extraction technique, therefore, delete the accuracy that the spectral absorption of these wavelength helps to improve model; , with optimization progressively spectrogram is carried out wavelength and select automatically to the optimization of collection of illustrative plates by OPUS software;

4) mensuration of sample index's property composition: with recording index component content in the calibration set sample with the corresponding detection method of sample in the conventional method;

5) set up Quantitative Analysis Model: the NIR spectrum of calibration set sample index property component content with it is combined, with partial least-square regression method (PLS method), principal component regression (PCR) or multiple linear regression (MLR) method are set up calibration model, wherein, the PLS method combines factorial analysis and regretional analysis, it is a kind of preferably method of more effect of using in the near-infrared spectrum analysis, partial least square method (PLS) method is a kind of full spectroscopic analysis methods, this method has made full use of the useful information under a plurality of wavelength, do not need painstakingly to select wavelength, and the noise of energy elimination raw data, improve signal to noise ratio (S/N ratio), be well suited among the NIR and use, the characteristics that PLS analyzes are:

1, can be from whole and part spectroscopic data information extraction; 2, data matrix decomposition and recurrence are combined into a step alternately, and the feature value vector that obtains is directly relevant with tested component or character, rather than relevant with the variable of variation maximum in the data matrix; If 3 calibration sets of selecting are representative, the PLS model is more sane; 4, can be used for complicated analytic system;

The rudimentary algorithm step of partial least square method is: at first, the concentration matrix is become loading matrix and gets sub matrix with the spectrum matrix decomposition, do principal component analysis (PCA) then, select suitable number of principal components, noise in filtering spectrum matrix and the concentration matrix, at last, utilize regretional analysis to obtain the correlation coefficient matrix, in the programming of actual calculation machine, usually a step is merged in spectrum matrix and the decomposition that gets sub matrix, and the concentration matrix information is introduced in the spectrum matrix decomposition process, before calculating a new component, concentration is got sub matrix and spectrum get sub matrix and exchange, make spectrum matrix major component related with the concentration matrix, this is the place that partial least square method is better than other analytic approach;

Its basic theories is: establish m blend sample, be made of n component, if record the absorption value of a sample L wavelength points, then can get the absorption value matrix A.The product that matrix A can be divided into two matrixes according to principal component analysis (PCA):

A＝TP+E

E is the inexplicable stochastic error matrix of system model;

Each concentration of component data formation concentration Matrix C also can be carried out same decomposition in each sample:

C＝UQ+F

F is the stochastic error matrix, and by Lambert-Beer's law as can be known, there are internal relations in A and C, so can set up following linear relationship:

U＝TB

B is a pair of angular moment battle array, so have:

C＝TB?Q+F

For unknown sample, by the matrix A of unknown sample _UnknownUtilize the relation of A=TP and the P that in aligning step, stores thereof, can calculate T _Unknown, then with aligning step in the B that stores obtain U, the Q by storage can obtain C _Unknown

6) forecast set (inspection set, as follows) sample, evaluation model: the near infrared collection of illustrative plates of scanning unknown sample, the Quantitative Analysis Model that the input of this collection of illustrative plates is set up, can dope the index component content (predicted value) of this unknown sample, compare with the value that records with standard method (actual value), come evaluation model, evaluating is as follows:

(1) coefficient R ²:

$R^2=1-\frac{\mathrm{\Σ}{(C_i-{\hat{C}}_i)}^2}{\mathrm{\Σ}{(C_i-C_m)}^2}$

The linear degree of this value representation predicted value and actual value relation;

(2) checking error mean square root (RMSEP);

$\mathrm{PMSEP}=\sqrt{\frac{\mathrm{\Σ}{({\hat{C}}_i-C_i)}^2}{m}}$

Deviation between this value representation predicted value and actual value;

(3) forecast set relative prediction residual (RSEP%);

$\mathrm{RSE}\%=\sqrt{\frac{\mathrm{\Σ}{({\hat{C}}_i-C_i)}^2}{\mathrm{\Σ}{C}_i^2}}\×100$

Represent the predicted value of forecast set and the relative deviation between actual value;

(4) cross validation error mean square root (RMSECV);

$\mathrm{RMSECV}=\sqrt{\frac{\mathrm{\Σ}{({\hat{C}}_i-C_i)}^2}{n-p}}$

Above-mentioned various in: C _i-traditional analysis method measured value;

-measure and the result of mathematical model prediction by NIR; N-sets up the calibration set sample number that model is used; The main cause subnumber that adopts in the p-calibration model; M-is used for the inspection set sample number of testing model;

If R ²Near 1, then the correlativity between the predicted value of calibration model and the standard control methods analyst value is strong more more, and RMSECV, RMSEP and RSE% value are littler, and then the precision of prediction of model is higher.

Embodiment 1:

Said Chinese medicine medicinal material is the root of large-flowered skullcap, and its quality overall evaluation is realized by following steps:

1) collection of the NIR diffuse reflection spectrum of radix scutellariae medicinal materials:

Instrument: VECTOR22-NIR type ft-nir spectrometer (German Brooker company), be furnished with PbS and InGaAs detecting device, external integrating sphere, sample spinner and solid fibre-optical probe; OPUS signals collecting and data processing software;

Sample: from 93 parts of radix scutellariae medicinal materials of national different cultivars, the different place of production, different collecting season, different concocting methods;

After 93 parts of radix scutellariae medicinal materials oven dry, cross 80 mesh sieves respectively, put into rotating cup, the medicinal powder of tiling 2～5cm thickness uses the collection of integrating sphere diffuse reflectance accessory, scanning times 64, resolution 8cm ^-1, the absorbance data form is: log1/R, spectral scan scope 12000～4000cm ^-1, spectroscopic data, get wherein 83 parts as the calibration set sample, this be forecast set for 10 increments, does to verify;

2) foundation of water and basis weight analytical model in the radix scutellariae medicinal materials

The selection of A, wave band and pre-service

, adopt progressively optimization that spectrogram is carried out wavelength and select automatically to the optimization of collection of illustrative plates by OPUS software, select 7502.2-4246.8cm ^-1As the modeling interval, this interval with moisture at the remarkable absorption peak in NIR district (5120 and 6850cm ^-1) the position overlaid, preprocessing procedures is selected polynary scatter correction method;

The mensuration of B, moisture

Adopt that " Chinese pharmacopoeia---oven drying method records in the root of large-flowered skullcap moisture and does reference, after the NIR spectral analysis, radix scutellariae medicinal materials in the rotating cup transferred in the measuring cup weigh, put into 105 ℃ baking oven 5 hours, in exsiccator, be cooled to room temperature (18-25 ℃) back weighing, again 105 ℃ of dryings 1 hour, cooling is weighed, and extremely double difference of weighing is no more than till the 5mg, according to the weight that subtracts mistake, calculate the percentage that contains moisture in the root of large-flowered skullcap sample;

The foundation of C, water and basis weight analytical model

The moisture data of 83 parts of calibration set samples that oven drying method is recorded combine with its NIR spectroscopic data, set up Quantitative Analysis Model with offset minimum binary (PLS) homing method;

D, verification of model

The NIR spectroscopic data of 10 parts of forecast set samples is imported this Quantitative Analysis Model, obtain predicted value, compare, estimate this Quantitative Analysis Model with the value that records with oven drying method (actual value);

Carry out internal chiasma checking RMSECV=0.458, R with the calibration samples collection ²=0.943, determine that best number of principal components is 10, the relative error between near infrared spectroscopy predicted value and the actual value is between ± 1.1%, and model tuning sum of errors predicated error is all less, the calibration model linear relationship of being set up is remarkable, and this model prediction precision is higher;

3) foundation of ethanol soluble extractives Quantitative Analysis Model in the radix scutellariae medicinal materials

The selection of A, wave band and pre-service

By OPUS software automatically to the optimization of collection of illustrative plates and a plurality of wave band to RMSECV and R ²The comparison of influence, adopt progressively optimization that spectrogram is carried out wavelength and select, select 11995.9-7498.4cm ^-1And 5450.2-4246.8cm ^-1As the modeling interval, preprocessing procedures is selected the vector normalization method;

Ethanol soluble extractives Determination on content in B, the radix scutellariae medicinal materials

" Chinese pharmacopoeia---hot dipping is measured in employing, 93 parts of root of large-flowered skullcap crude drugs are got respectively about the root of large-flowered skullcap crude drug powder 2g of each part, put in the conical flask of 100～250ml, the adding mass concentration is 70% ethanol 50ml, claims to decide weight, after leaving standstill 1 hour, the continuous backflow condenser pipe is heated to boiling, and keeps little and boiled 1 hour, after the cooling, take off conical flask, close plug claims to decide weight again, with mass concentration is that 70% ethanol is supplied the weight that subtracts mistake, shake up, filter, measure filtrate 25ml with dry filter, put in the dry evaporating dish, behind evaporate to dryness in the water-bath,, put and cool off 30min in the exsiccator in 105 ℃ of dry 3h, claim to decide weight rapidly, calculate the content results such as the table 3 of ethanol soluble extractives in the root of large-flowered skullcap sample with dry product:

Ethanol soluble extractives in 93 parts of radix scutellariae medicinal materials of table 3

The ethanol soluble extractives content distribution is more even in the 93 batches of radix scutellariae medicinal materials as can be seen from Table 3, and ethanol soluble extractives content is 34.42%-56.00%, meets the basic demand of near infrared modeling.

The foundation of ethanol soluble extractives Quantitative Analysis Model in C, the radix scutellariae medicinal materials

The ethanol soluble extractives content data of the radix scutellariae medicinal materials of 83 parts of calibration set samples that hot dipping is recorded combines with its NIR spectroscopic data, sets up Quantitative Analysis Model with offset minimum binary (PLS) homing method;

D, verification of model

The NIR spectroscopic data of 10 parts of inspection set samples is imported this Quantitative Analysis Model, obtain predicted value, compare with the value that records with oven drying method (actual value), can obtain the forecast set mean absolute error as calculated is 0.2814, and average relative error is 2.11%.The foundation of model is successful as can be seen, and it is comparatively accurate to predict the outcome;

4) foundation of scutelloside Quantitative Analysis Model in the root of large-flowered skullcap crude drug

The selection of A, wave band and pre-service

By OPUS software automatically to the optimization of collection of illustrative plates and a plurality of wave band to RMSECV and R ²The comparison of influence, adopt progressively optimization that spectrogram is carried out wavelength and select, select 7502.2-4246.8cm ^-1As the modeling interval, preprocessing procedures is selected polynary scatter correction method.

The mensuration of B, content of baicalin

With " the high-efficient liquid phase technique method of Chinese pharmacopoeia version in 2005 is carried out assay to radix scutellariae medicinal materials, and chromatographic condition and system suitability test are filling agent with octadecylsilane chemically bonded silica; Methanol-water-phosphoric acid (47: 53: 0.2) is a moving phase; The detection wavelength is 280nm, number of theoretical plate calculates by the scutelloside peak should be not less than 2500, reference substance (is standard items, we obtain this standard items by purchase, such as we will survey the content of baicalin in the radix scutellariae medicinal materials, just the scutelloside standard items must be arranged earlier, we use the content of baicalin in the high-performance liquid chromatogram determination medicinal material, the solution that the standard items that configure just must be arranged) preparation of solution is: take by weighing at 4 hours scutelloside reference substance of 60 ℃ of drying under reduced pressure an amount of, add methyl alcohol and make the solution that every 1ml contains 60 μ g, promptly; The preparation of root of large-flowered skullcap sample solution is: get root of large-flowered skullcap crude drug powder 0.3g, add mass concentration and be 70% ethanol 40ml, reflux 3 hours, put coldly, filter, filtrate is put in the 100ml volumetric flask, with mass concentration is 70% ethanol gradation washing container and residue, and washing lotion is filtered in the same volumetric flask, adds mass concentration and be 70% ethanol to 100ml, shake up, measure 1ml, put in the 10ml volumetric flask, add methyl alcohol to 10ml, shake up, promptly; Determination method is: draw each 10 μ l of reference substance solution and root of large-flowered skullcap sample solution respectively, inject liquid chromatograph, measure, promptly; Measurement result such as table 5:

The content of baicalin of table 5 93 duplicate samples

Content of baicalin distribution uniform in the 93 batches of root of large-flowered skullcap crude drugs as can be seen from Table 5, content of baicalin is 3.06%-20.19%, meets the basic demand of near infrared modeling.

The foundation of C, scutelloside Quantitative Analysis Model

The content of baicalin data of 83 parts of calibration set samples that oven drying method is recorded combine with its NIR spectroscopic data, set up Quantitative Analysis Model with offset minimum binary (PLS) homing method;

D, verification of model

The NIR spectroscopic data of 10 parts of inspection set samples is imported this Quantitative Analysis Model, obtain predicted value, compare, estimate this Quantitative Analysis Model, carry out internal chiasma checking RMSECV=1.29, R with the calibration samples collection with the value that records with oven drying method (actual value) ²=0.906, determine that best number of principal components is 10, the relative error between near infrared spectroscopy predicted value and the actual value is between 2.8%-2.2%, and model tuning sum of errors predicated error is all less.

Embodiment 2:

Said Chinese medicine medicinal material is the capsule of weeping forsythia, and its quality overall evaluation is realized by following steps:

The collection of the NIR diffuse reflection spectrum of capsule of weeping forsythia medicinal material:

Instrument: Nicolet6700 type ft-nir spectrometer, be furnished with InGaAs detecting device, external integrating sphere, sample spinner and solid fibre-optical probe; OMNIC spectra collection software and TQ7.2 analysis software;

Sample: 100 parts in wild capsule of weeping forsythia medicinal material sample containing each species diversity from ground different cultivars such as Henan, Shanxi, Shaanxi, Hebei, Hubei, Shandong, different collecting season, different concocting methods etc.;

Sample is dried, cross 24 mesh sieves after the pulverizing respectively, put into rotating cup, the medicinal powder of tiling 2-5cm thickness uses the collection of integrating sphere diffuse reflectance accessory, scanning times 32, resolution 8cm ^-1, temperature: (20 ± 0.3) ℃, relative humidity: 35%, spectral scan scope 12000～4000cm ^-1

1), the foundation of capsule of weeping forsythia medicinal material water and basis weight analytical model

The selection of A, wave band and pre-service

By of the optimization of TQ data processing software, with optimization progressively spectrogram is carried out wavelength and select collection of illustrative plates; Selected range of wavelengths is 7502.2-5046.8cm ^-1, the correlation of this SPECTRAL REGION is best, and the spectrum pre-service is MSC (polynary scatter correction)+First Derivative (first order derivative) method;

B, " Chinese pharmacopoeia---toluene method records moisture in the capsule of weeping forsythia medicinal material in the mensuration employing of moisture, and select wherein 65 increments this as the calibration set sample, 22 increments this as the forecast set sample, get capsule of weeping forsythia sample 200g, adopt the distiller condenser device, capsule of weeping forsythia 200g is put in the trial jar, add about toluene 200ml, add beaded glass number in case of necessity, the instrument each several part is connected, autocondensation pipe top adds toluene, to the narrow thin part that is full of test tube, the A bottle put in the electric jacket or with other proper method slowly heat, when treating that toluene comes to life, regulate temperature, make and distillate 2 p.s., treat that moisture distillates fully, condenser pipe is inner earlier with the toluene flushing, again with full long brush or other suitable methods of dipping in toluene, the toluene that adheres on the tube wall is pushed, continue distillation 5 minutes, put and be chilled to room temperature 10-30 ℃, stripping assembly, if any water adhesion on the tube wall of B pipe, the available copper wire that dips in toluene pushes, and places, and makes moisture separate fully with toluene that (it is a small amount of to add the methylenum careuleum powder, make water dye blueness, observe so that separate), the water yield is read in inspection, and is calculated to be the percentage that contains moisture in the capsule of weeping forsythia sample;

The foundation of C, water and basis weight analytical model

The moisture data that toluene method is recorded combine with its NIR spectroscopic data, set up Quantitative Analysis Model with offset minimum binary (PLS) homing method;

D, verification of model

The NIR spectroscopic data of 22 parts of forecast set samples is imported this Quantitative Analysis Model, obtain predicted value, compare, estimate this Quantitative Analysis Model, carry out internal chiasma checking RMSECV=1.66, R with the calibration samples collection with the value that records with toluene method (actual value) ²=0.92, can draw, the calibration model linear relationship of being set up is remarkable, and model tuning sum of errors predicated error is all less, and this model prediction precision is higher;

2), the foundation of capsule of weeping forsythia medicinal material extract Quantitative Analysis Model

The selection of A, wave band and pre-service

By of the optimization of TQ data processing software to collection of illustrative plates, with optimization progressively spectrogram is carried out wavelength and select, selected range of wavelengths is 4516.19-7058.19cm ^-1, the correlation of this SPECTRAL REGION is best, and the spectrum pre-service is polynary scatter correction method;

B, capsule of weeping forsythia medicinal material ethanol soluble extractives Determination on content

Adopt " Chinese pharmacopoeia---cold-maceration records ethanol soluble extractives content in the capsule of weeping forsythia medicinal material, and select wherein 70 increments this as the calibration set sample, 24 increments this as the forecast set sample;

Cold-maceration: get the capsule of weeping forsythia sample 4g after the pulverizing, put in the conical flask of 250-300ml, adding mass concentration is 65% ethanol 100ml, close plug, cold soaking, jolting constantly in preceding 6 hours, left standstill again 18 hours, and filtered rapidly, draw filtrate 20ml with dry filter, put in the dry evaporating dish, behind evaporate to dryness in the water-bath, in 105 ℃ of dryings 3 hours, move in the exsiccator, cooled off 30 minutes, and claimed to decide weight rapidly, calculate the percentage of ethanol soluble extractives in the capsule of weeping forsythia sample;

The foundation of C, capsule of weeping forsythia medicinal material ethanol soluble extractives Quantitative Analysis Model

The ethanol soluble extractives content data that cold-maceration is recorded combines with its NIR spectroscopic data, sets up Quantitative Analysis Model with offset minimum binary (PLS) homing method;

D, verification of model

The NIR spectroscopic data of 24 parts of inspection set samples is imported this Quantitative Analysis Model, obtain predicted value, compare with the value that records with cold-maceration (actual value), with the ratio between NIR predicted value and the pharmacopeia method measurement result as the prediction recovery, average recovery rate be 101.05% (recovery in the notion of Chinese medicine about 100%, be normal at this between the 98-102%), for given level of significance 0.05, t (0.05,23)=2.069, the paired t test value of gained is less than 2.069, the analysis result that two kinds of methods are described does not have significant difference, this model of above presentation of results can accurately be predicted the capsule of weeping forsythia medicinal material extract content of its coverage by checking;

3), the foundation of capsule of weeping forsythia medicinal material forsythin Quantitative Analysis Model

The selection of A, wave band and pre-service

By the optimization of TQ data processing software to collection of illustrative plates, selected range of wavelengths is 4000-10000cm ^-1, the correlation of this SPECTRAL REGION is best, and the spectrum pre-service is the first order derivative method;

B, forsythin Determination on content

Adopt " Chinese pharmacopoeia---cold-maceration records forsythin content in the capsule of weeping forsythia medicinal material, and select wherein 65 increments this as the calibration set sample, 26 increments this as the forecast set sample;

Chromatographic condition is filling agent with the octadecylsilane chemically bonded silica; Moving phase is acetonitrile-water (23: 77); Flow velocity 1.0ml/min; The detection wavelength is 277nm; 30 ℃ of column temperatures; The external standard peak area method is quantitative;

The preparation of reference substance solution: it is an amount of to take by weighing the forsythin reference substance, adds methyl alcohol and makes the solution that the forsythin mass concentration is 0.2mg/ml.

The preparation of capsule of weeping forsythia sample solution: get capsule of weeping forsythia medicinal powder 1g, put in the tool plug conical flask, add methyl alcohol 15ml, claim to decide weight, dipping spent the night sonicated 25 minutes 12-14 hour, put coldly, claim again to decide weight, supply the weight that subtracts mistake with methyl alcohol, shake up, filter, measure filtrate 5ml, steam near and do, add neutral alumina 0.5g and mix thoroughly, add and put neutral alumina post (100～120 orders, 1g on the internal diameter 1～1.5cm), is 70% ethanol 80ml wash-out with mass concentration, collect eluent, be concentrated into driedly, the residue mass concentration is 50% dissolve with methanol, be transferred in the 5ml volumetric flask, and be diluted to 5ml, and shake up, filter, get filtrate, promptly;

Determination method: draw each 10 μ l of reference substance solution and capsule of weeping forsythia sample solution respectively, inject liquid chromatograph and measure, calculate the percentage of forsythin in the capsule of weeping forsythia sample;

The foundation of C, forsythin Quantitative Analysis Model

The forsythin content data that cold-maceration is recorded combines with its NIR spectroscopic data, sets up Quantitative Analysis Model with offset minimum binary (PLS) homing method;

D, verification of model

The NIR spectroscopic data of 26 parts of forecast set samples is imported this Quantitative Analysis Model, obtain predicted value, compare with the value that records with cold-maceration (actual value), with the ratio between NIR predicted value and the pharmacopeia method measurement result as the prediction recovery, getting average recovery rate is 104.72%, for given level of significance 0.05, t (0.05,25)=2.060, the paired t test value of gained is less than 2.060, the analysis result that two kinds of methods are described does not have significant difference, and this model of above presentation of results can accurately be predicted the capsule of weeping forsythia medicinal material forsythin content of its coverage by checking.

Embodiment 3:

Said Chinese medicine medicinal material is a Chinese yam, and its near-infrared spectrum technique is realized by following steps the quality overall evaluation of Chinese yam:

The collection of the NIR diffuse reflection spectrum of Chinese yam medicinal material:

Instrument: VECTOR22-NIR type ft-nir spectrometer (German Brooker company), be furnished with PbS and InGaAs detecting device, external integrating sphere, sample spinner and solid fibre-optical probe; OPUS signals collecting and data processing software;

Sample: from 115 parts of Chinese yam medicinal material samples of national different cultivars, the different place of production, different collecting seasons;

After 115 parts of Chinese yam medicinal material oven dry, pulverized 100 mesh sieves respectively, put into rotating cup, the medicinal powder of tiling 2～5cm thickness uses the collection of integrating sphere diffuse reflectance accessory, scanning times 64, resolution 8cm ^-1, the absorbance data form is: log1/R, spectral scan scope 12000～4000cm ^-1, spectroscopic data, get wherein 105 parts as the calibration set sample, this be forecast set for 10 increments, does to verify;

1) foundation of Chinese yam medicinal material polysaccharide Quantitative Analysis Model

The selection of A, wave band and pre-service

By of the optimization of TQ data processing software to collection of illustrative plates, with optimization progressively spectrogram is carried out wavelength and select, selected range of wavelengths is 7513.8～4597.8cm ^-1, the correlation of this SPECTRAL REGION is best, and the spectrum pre-service is First Derivative (first order derivative method)+Vector Normalization (vector normalization method);

B, the measurement of the polysaccharide content of Chinese yam medicinal material

Get coarse yam powder (crossing sieve No. two) 20g, put in the apparatus,Soxhlet's, add 90 ℃ of backflow 4h of 250ml sherwood oil, after the taking-up coarse yam powder volatilizes, the adding mass concentration is 80% twice (each two hours) of ethanol 200ml (90 ℃) backflow, to remove small-molecule substances such as monose, polyphenol, compound sugar and glycoside; After coarse yam powder volatilized solvent, add the water of 200ml, 80 ℃ of water-baths are extracted twice, the each extraction 4 hours merges extract, is concentrated into 50ml with the rotary evaporator vacuum, after removing albumen with the sevage method, vacuum is concentrated into 20ml once more, adds 95% ethanol, makes the concentration of alcohol in the concentrate of 20ml reach 80%, standing over night 12-14 hour, filter, residue successively washs successively with absolute ethyl alcohol, ether, acetone, and is dry below 60 ℃;

Take by weighing Chinese yam polysaccharide 10mg, be dissolved in the 100ml volumetric flask, be diluted to 100ml, shake up, it is an amount of to draw above-mentioned solution, puts in the 25ml volumetric flask, and adding distil water promptly gets the Chinese yam sample solution to 25ml, records polyoses content with toluene method;

The foundation of C, Chinese yam medicinal material polysaccharide Quantitative Analysis Model

The polyoses content data that toluene method is recorded combine with its NIR spectroscopic data, set up Quantitative Analysis Model with offset minimum binary (PLS) homing method;

D, verification of model

The NIR spectroscopic data of 22 parts of forecast set samples is imported this Quantitative Analysis Model, obtain predicted value, compare, estimate this Quantitative Analysis Model with the value that records with toluene method (actual value).Carry out internal chiasma checking RMSECV=1.18, R with the calibration samples collection ²=0.93, can draw, the calibration model linear relationship of being set up is remarkable, and model tuning sum of errors predicated error is all less, and this model prediction precision is higher;

2), the foundation of Chinese yam medicinal material allantoin Quantitative Analysis Model

The selection of A, wave band and pre-service

By of the optimization of TQ data processing software, with optimization progressively spectrogram is carried out wavelength and select collection of illustrative plates.Selected range of wavelengths is 7513.8～6094.4cm ^-1With 5461.8～4242.9cm ^-1, the correlation of this SPECTRAL REGION is best, and the spectrum pre-service is polynary scatter correction method;

B, Chinese yam medicinal material allantoin Determination on content

Adopt high-efficient liquid phase technique to record allantoin content in the Chinese yam medicinal material, and select wherein 105 increments this as the calibration set sample, 10 increments this as the forecast set sample, got 100 mesh sieves, the Chinese yam fine powder 1g of dry 3h below 60 ℃, put in the ground conical flask, adding mass concentration is 50% ethanol 50ml, weighs, ultrasonic Extraction 30min, it is heavy to place room temperature (being 18-25 ℃) back benefit, filters, and gets filtrate 25ml in evaporating dish, water-bath concentrates to the greatest extent and does, water is dissolved to 10ml, and it is centrifugal in right amount to get each Chinese yam sample solution, gets supernatant sample introduction 5 μ l respectively.Measure peak area by above-mentioned chromatographic condition, with the content of allantoin in the external standard method calculation sample;

The foundation of C, Chinese yam medicinal material allantoin Quantitative Analysis Model

The allantoin content data that high performance liquid chromatography is recorded combines with its NIR spectroscopic data, sets up Quantitative Analysis Model with offset minimum binary (PLS) homing method;

D, verification of model

The NIR spectroscopic data of 10 parts of forecast set samples is imported this Quantitative Analysis Model, obtain predicted value, compare with the value that records with high performance liquid chromatography (actual value), the related coefficient of allantoin measured value and actual value is 0.9598, and RMSEP is 0.0329.The measured value of allantoin and the absolute error between the actual value are between ± 0.1% in the inspection set, and what actual value was relatively concentrated is distributed near the center line, illustrates that measured value and actual value are more identical, and the foundation of model is comparatively successful as can be seen;

3), the foundation of Chinese yam medicinal material water extraction Quantitative Analysis Model

The selection of A, wave band and pre-service

By of the optimization of TQ data processing software to collection of illustrative plates, with optimization progressively spectrogram is carried out wavelength and select, selected range of wavelengths is 7513.8～4242.9cm ^-1, the correlation of this SPECTRAL REGION is best, and the spectrum pre-service is First Derivative (first order derivative method)+Vector Normalization (vector normalization method);

B, Chinese yam medicinal material water extraction Determination on content

Adopt high-efficient liquid phase technique to record water extraction content in the Chinese yam medicinal material, and select wherein 105 increments this as the calibration set sample, 10 increments this as the forecast set sample, adopt hot dipping: get the Chinese yam sample 4g of 60 ℃ of dryings, put in the conical flask of 250ml, add water 100ml, close plug, claim decide weight, leave standstill 1 hour after, the connection reflux condensing tube, be heated to boiling, and keep little and boiled 1 hour, after the cooling, take off conical flask, close plug, claim to decide weight, water is supplied the weight that subtracts mistake, shakes up again, filter with dry filter, measure filtrate 25ml, put in the dry evaporating dish, behind evaporate to dryness in the water-bath, in 105 ℃ of dryings 3 hours, put in the exsiccator and cooled off 30 minutes, claim to decide weight rapidly, calculate the content of water-soluble extractives in the Chinese yam sample with dry product;

The foundation of C, Chinese yam medicinal material water extraction Quantitative Analysis Model

The water extraction content data that cold-maceration is recorded combines with its NIR spectroscopic data, sets up Quantitative Analysis Model with offset minimum binary (PLS) homing method;

D, verification of model

The NIR spectroscopic data of 10 parts of forecast set samples is imported this Quantitative Analysis Model, obtain predicted value, compare, estimate this Quantitative Analysis Model with the value that records with cold-maceration (actual value).Carry out internal chiasma checking RMSECV=1.23, R with the calibration samples collection ²=0.94, can draw, the calibration model linear relationship of being set up is remarkable, model tuning sum of errors predicated error is all less, this model prediction precision is higher, determines that best number of principal components is 12, and the absolute error between near infrared spectroscopy measured value and the actual value is between ± 3%.

The present invention, show through above result of study, by setting up polynary calibration model, the NIR spectroscopic analysis methods can effectively detect the moisture in the Chinese crude drug, near infrared light (NIR) is meant the electromagnetic wave of wavelength in 780～2526nm scope, that write down mainly is hydrogeneous radicals X-H (X=C, N, O) Zhen Dong frequency multiplication and sum of fundamental frequencies absorb, different groups or absorbing wavelength and the intensity of same group in the different chemical environment all have significant difference, effective constituent almost all has absorption in this zone in the Chinese crude drug, therefore, NIR is very suitable for the quality testing of Chinese crude drug, this method only needs simple sample preparation, compares with standard method, can save a large amount of analysis times and cost, being applicable to the fast detecting of moisture in most of Chinese crude drug, is a kind of convenience, fast, harmless green analytical technology.

Claims (4)

1, a kind of method of utilizing near-infrared spectrum technique to estimate Chinese medicine medicinal material overall quality is characterized in that realized by following steps: at first set up calibration model, method is:

1) collect the calibration set sample: the medicinal material of promptly collecting each species diversity is as the calibration set sample, and is the traditional Chinese medicinal material samples crushing screening of collecting, standby;

2) gather sample spectrum: use near infrared spectrometer, the near-infrared diffuse reflection spectrum of acquisition correction collection sample obtains spectroscopic data, and sample is divided into calibration set and forecast set two classes;

3) sample spectrum pre-service: method is:

Polynary scatter correction MSC: this scattering effect of eliminating near infrared spectrum; Or use derivative method, and the first order derivative in the derivative method is eliminated the skew of spectrum baseline, and second derivative is eliminated the linear tilt of baseline, and after the broad peak in the former spectrum was handled through second derivative, crest was sharpened, determines the position at peak, improves signal to noise ratio (S/N ratio); By OPUS software spectrogram being carried out wavelength selects;

4) mensuration of sample index's property composition: just collecting index component content in the sample on the school with recording with the corresponding detection method of sample in the conventional method;

5) set up Quantitative Analysis Model: the NIR spectrum of calibration set sample index property component content with it is combined, with partial least-square regression method full spectrum is analyzed, set up calibration model, method is: at first, the concentration matrix is become loading matrix and gets sub matrix with the spectrum matrix decomposition, do principal component analysis (PCA) then, noise in filtering spectrum matrix and the concentration matrix, at last, utilize regretional analysis to obtain the correlation coefficient matrix, and with in the concentration matrix information introducing spectrum matrix decomposition process, before calculating a new component, concentration is got sub matrix and spectrum get sub matrix and exchange, make spectrum matrix major component related with the concentration matrix, correlation is: wherein A is a spectrum matrix major component absorption value matrix

A＝TP+E

E is the inexplicable stochastic error matrix of system model;

Each concentration of component data constitutes the concentration Matrix C in each sample:

C＝UQ+F

F is the stochastic error matrix:

U＝TB

B is a pair of angular moment battle array:

C＝TB?Q+F

For unknown sample, by the matrix A of unknown sample _UnknownUtilize the relation of A=TP and the P that in aligning step, stores thereof, can calculate T _Unknown, then with aligning step in the B that stores obtain U, the Q by storage can obtain C the unknown;

6) forecast set sample, evaluation model: the near infrared collection of illustrative plates of scanning unknown sample, the Quantitative Analysis Model that the input of this collection of illustrative plates is set up, can dope the index component content of this unknown sample, compare with the value that records with standard method, come evaluation model, evaluating is as follows:

(1) coefficient R ²:

$R^2=1-\frac{\mathrm{\Σ}{(C_1-{\hat{C}}_1)}^2}{\mathrm{\Σ}{(C_1-C_m)}^2}$

The linear degree of this value representation predicted value and actual value relation;

(2) checking error mean square root (RMSEP);

$\mathrm{RMSEP}=\sqrt{\frac{\mathrm{\Σ}{({\hat{C}}_1-C_1)}^2}{m}}$

Deviation between this value representation predicted value and actual value;

(3) forecast set relative prediction residual (RSEP%);

$\mathrm{RSE}\%=\sqrt{\frac{\mathrm{\Σ}{({\hat{C}}_1-C_1)}^2}{\mathrm{\Σ}{C}_1^2}}\×100$

Represent the predicted value of forecast set and the relative deviation between actual value;

(4) cross validation error mean square root (RMSECV);

$\mathrm{RMSECV}=\sqrt{\frac{\mathrm{\Σ}{({\hat{C}}_1-C_1)}^2}{n-p}}$

Above-mentioned various in: C _i-traditional analysis method measured value;

-measure and the result of mathematical model prediction by NIR; N-sets up the calibration set sample number that model is used; The main cause subnumber that adopts in the p-calibration model; M-is used for the inspection set sample number of testing model;

If R ²Near 1, then the correlativity between the predicted value of calibration model and the standard control methods analyst value is strong more more, and RMSECV, RMSEP and RSE% value are littler, and then the precision of prediction of model is higher.

2, the method for utilizing near-infrared spectrum technique to estimate Chinese medicine medicinal material overall quality according to claim 1 is characterized in that said Chinese medicine medicinal material is the root of large-flowered skullcap, and its quality overall evaluation is realized by following steps:

1) collection of the NIR diffuse reflection spectrum of radix scutellariae medicinal materials:

By near infrared spectrometer, be furnished with PbS and InGaAs detecting device, external integrating sphere, sample spinner and solid fibre-optical probe and OPUS signals collecting and data processing software, 93 parts of radix scutellariae medicinal materials from national different cultivars, the different place of production, different collecting season, different concocting methods, scan, get spectroscopic data, method is:

After the oven dry of 93 parts of radix scutellariae medicinal materials, cross 80 mesh sieves respectively, put into rotating cup, the medicinal powder of tiling 2～5cm thickness uses the collection of integrating sphere diffuse reflectance accessory, spectroscopic data, get wherein 83 parts as the calibration set sample, this is forecast set for 10 increments;

2) set up water and basis weight analytical model in the radix scutellariae medicinal materials:

The selection of A, wave band and pre-service

By OPUS software spectrogram is carried out wavelength and select, select 7502.2-4246.8cm ^-1As the modeling interval, this is interval with remarkable absorption peak 5120 and the 6850cm of moisture in the NIR district ^-1Position overlaid, preprocessing procedures are selected polynary scatter correction method;

The mensuration of B, moisture

Adopt that " Chinese pharmacopoeia---oven drying method records in the root of large-flowered skullcap moisture and does reference, after the NIR spectral analysis, radix scutellariae medicinal materials in the rotating cup transferred in the measuring cup weighs, put into 105 ℃ baking oven 5 hours, in exsiccator, be cooled to room temperature 18-25 ℃ after weighing, again 105 ℃ of dryings 1 hour, cooling is weighed, and extremely double difference of weighing is no more than till the 5mg, according to the weight that subtracts mistake, calculate the percentage that contains moisture in the root of large-flowered skullcap sample;

C, set up the water and basis weight analytical model:

The moisture data of 83 parts of calibration set samples that oven drying method is recorded combine with its NIR spectroscopic data, set up Quantitative Analysis Model with partial least-square regression method;

D, verification of model:

The NIR spectroscopic data of 10 parts of forecast set samples is imported this Quantitative Analysis Model, obtain predicted value, compare, estimate this Quantitative Analysis Model with the value that records with oven drying method;

Carry out internal chiasma checking RMSECV=0.458, R with the calibration samples collection ²=0.943, determine that best number of principal components is 10, the relative error between near infrared spectroscopy predicted value and the actual value is between ± 1.1%;

3) set up ethanol soluble extractives Quantitative Analysis Model in the radix scutellariae medicinal materials

The selection of A, wave band and pre-service

By OPUS software a plurality of wave bands of collection of illustrative plates are carried out RMSECV and R ²The comparison of influence, spectrogram is carried out wavelength selects, select 11995.9-7498.4cm ^-1And 5450.2-4246.8cm ^-1As the modeling interval, preprocessing procedures is selected the vector normalization method;

Ethanol soluble extractives Determination on content in B, the radix scutellariae medicinal materials:

" the Chinese pharmacopoeia hot dipping is measured in employing, 93 parts of root of large-flowered skullcap crude drugs are got respectively about the root of large-flowered skullcap crude drug powder 2g of each part, put in the conical flask of 100～250ml, add mass concentration and be 70% ethanol 50ml, claim to decide weight, leave standstill 1 hour after, the continuous backflow condenser pipe, be heated to boiling, and keep little and boiled 1 hour, after the cooling, take off conical flask, close plug claims to decide weight again, is that 70% ethanol is supplied the weight that subtracts mistake with mass concentration, shake up, filter with dry filter, measure filtrate 25ml, put in the dry evaporating dish, behind evaporate to dryness in the water-bath, in 105 ℃ of dry 3h, to put and cool off 30min in the exsiccator, the content that calculates ethanol soluble extractives in the root of large-flowered skullcap sample with dry product is 34.42%-56.00%;

C, set up ethanol soluble extractives Quantitative Analysis Model in the radix scutellariae medicinal materials:

The ethanol soluble extractives content data of the radix scutellariae medicinal materials of 83 parts of calibration set samples that hot dipping is recorded combines with its NIR spectroscopic data, sets up Quantitative Analysis Model with partial least-square regression method;

D, verification of model

The NIR spectroscopic data of 10 parts of inspection set samples is imported this Quantitative Analysis Model, obtain predicted value, compare with the value that records with oven drying method, calculating the forecast set mean absolute error is 0.2814, and average relative error is 2.11%;

4) set up scutelloside Quantitative Analysis Model in the root of large-flowered skullcap crude drug:

The selection of A, wave band and pre-service

By OPUS software a plurality of wave bands of collection of illustrative plates are carried out RMSECV and R ²The comparison of influence, spectrogram is carried out wavelength selects, select 7502.2-4246.8cm ^-1As the modeling interval, preprocessing procedures is selected polynary scatter correction method;

The mensuration of B, content of baicalin

With " the high-efficient liquid phase technique method of Chinese pharmacopoeia is carried out assay to radix scutellariae medicinal materials, and chromatographic condition and system suitability test are filling agent with octadecylsilane chemically bonded silica; Methyl alcohol: water: the phosphoric acid volume ratio is 47: 53: 0.2 a moving phase; The detection wavelength is 280nm, and the scutelloside peak is not less than 2500, and the preparation of reference substance solution is: take by weighing at 4 hours scutelloside reference substance of 60 ℃ of drying under reduced pressure in right amount, add methyl alcohol and make the solution that every 1ml contains 60 μ g, promptly; The preparation of root of large-flowered skullcap sample solution is: get root of large-flowered skullcap crude drug powder 0.3g, add mass concentration and be 70% ethanol 40ml, reflux 3 hours, put coldly, filter, filtrate is put in the 100ml volumetric flask, with mass concentration is 70% ethanol gradation washing container and residue, and washing lotion is filtered in the same volumetric flask, adds mass concentration and be 70% ethanol to 100ml, shake up, measure 1ml, put in the 10ml volumetric flask, add methyl alcohol to 10ml, shake up, promptly; Determination method is: draw each 10 μ l of reference substance solution and root of large-flowered skullcap sample solution respectively, inject liquid chromatograph, measure, promptly getting content of baicalin is 3.06%-20.19%;

C, set up the scutelloside Quantitative Analysis Model:

The content of baicalin data of 83 parts of calibration set samples that oven drying method is recorded combine with its NIR spectroscopic data, set up Quantitative Analysis Model with partial least-square regression method;

D, verification of model

The NIR spectroscopic data of 10 parts of inspection set samples is imported this Quantitative Analysis Model, obtain predicted value, compare, estimate this Quantitative Analysis Model, carry out internal chiasma checking RMSECV=1.29, R with the calibration samples collection with the value that records with oven drying method ²=0.906, determine that best number of principal components is 10, the relative error between near infrared spectroscopy predicted value and the actual value is between 2.8%-2.2%.

3, the method for utilizing near-infrared spectrum technique to estimate Chinese medicine medicinal material overall quality according to claim 1 is characterized in that said Chinese medicine medicinal material is the capsule of weeping forsythia, and its quality overall evaluation is realized by following steps:

The collection of the NIR diffuse reflection spectrum of capsule of weeping forsythia medicinal material: collect from differently, 100 parts in capsule of weeping forsythia medicinal material samples such as different cultivars, different collecting season, different concocting methods; Sample is dried, pulverize the back respectively and cross 24 mesh sieves, put into rotating cup, the medicinal powder of tiling 2-5cm thickness uses integrating sphere diffuse reflection and spectra collection software, scans with near infrared spectrometer, temperature: 20 ± 0.3 ℃, relative humidity: 35%, get spectroscopic data;

1), set up capsule of weeping forsythia medicinal material water and basis weight analytical model:

The selection of A, wave band and pre-service

By the TQ data processing software spectrum is carried out wavelength and select, range of wavelengths is 7502.2-5046.8cm ^-1

B, " the Chinese pharmacopoeia toluene method records moisture in the capsule of weeping forsythia medicinal material in the mensuration employing of moisture, and select wherein 65 increments this as the calibration set sample, 22 increments this as the forecast set sample, capsule of weeping forsythia 200g adopts the distiller condenser device, and capsule of weeping forsythia 200g is put in the trial jar, add about toluene 200ml, autocondensation pipe top adds toluene, to the narrow thin part that is full of test tube, trial jar is put slowly heating in the electric jacket, when treating that toluene comes to life, regulate temperature, make to distillate 2 p.s., treat that moisture distillates fully, flushing, continue distillation 5 minutes, put and be chilled to 10-30 ℃, moisture is separated fully with toluene;

C, set up the water and basis weight analytical model

The moisture data that toluene method is recorded combine with its NIR spectroscopic data, set up Quantitative Analysis Model with partial least-square regression method;

D, verification of model

The NIR spectroscopic data of 22 parts of forecast set samples is imported this Quantitative Analysis Model, obtain predicted value, compare, estimate this Quantitative Analysis Model, carry out internal chiasma checking RMSECV=1.66, R with the calibration samples collection with the value that records with toluene method ²=0.92;

2), set up capsule of weeping forsythia medicinal material extract Quantitative Analysis Model:

The selection of A, wave band and pre-service

By the TQ data processing software spectrogram is carried out wavelength and select, range of wavelengths is 4516.19-7058.19cm ^-1

B, capsule of weeping forsythia medicinal material ethanol soluble extractives Determination on content adopt " the Chinese pharmacopoeia cold-maceration records ethanol soluble extractives content in the capsule of weeping forsythia medicinal material, and select wherein 70 increments this as the calibration set sample, 24 increments this as the forecast set sample;

Said cold-maceration is: get the capsule of weeping forsythia medicinal material 4g after capsule of weeping forsythia sample is pulverized, put in the conical flask of 250-300ml, adding mass concentration is 65% ethanol 100ml, close plug, cold soaking, jolting constantly in preceding 6 hours, left standstill again 18 hours, and filtered rapidly, draw filtrate 20ml with dry filter, put in the dry evaporating dish, behind evaporate to dryness in the water-bath, in 105 ℃ of dryings 3 hours, move in the exsiccator, cooled off 30 minutes, and claimed to decide weight rapidly, calculate the percentage of ethanol soluble extractives in the capsule of weeping forsythia sample;

C, set up capsule of weeping forsythia medicinal material ethanol soluble extractives Quantitative Analysis Model:

The ethanol soluble extractives content data that cold-maceration is recorded combines with its NIR spectroscopic data, sets up Quantitative Analysis Model with offset minimum binary (PLS) homing method;

D, verification of model:

The NIR spectroscopic data of 24 parts of inspection set samples is imported this Quantitative Analysis Model, obtain predicted value, compare with the value that records with cold-maceration, as the prediction recovery, getting average recovery rate is 101.05% capsule of weeping forsythia medicinal material extract content with accurate its coverage of prediction with the ratio between NIR predicted value and the pharmacopeia method measurement result;

3), set up capsule of weeping forsythia medicinal material forsythin Quantitative Analysis Model

The selection of A, wave band and pre-service

Selecting range of wavelengths by the TQ data processing software is 4000-10000cm ^-1

B, forsythin Determination on content

Adopt " the Chinese pharmacopoeia cold-maceration records forsythin content in the capsule of weeping forsythia medicinal material, and select wherein 65 increments this as the calibration set sample, 26 increments this as the forecast set sample;

Chromatographic condition is a filling agent with the octadecylsilane chemically bonded silica; Moving phase is acetonitrile: its volume ratio of water is 23: 77; Flow velocity 1.0ml/min; The detection wavelength is 277nm; 30 ℃ of column temperatures;

The preparation of reference substance solution: it is an amount of to take by weighing the forsythin reference substance, adds methyl alcohol and makes the solution that the forsythin mass concentration is 0.2mg/ml;

The preparation of capsule of weeping forsythia sample solution: get capsule of weeping forsythia medicinal powder 1g, put in the tool plug conical flask, add methyl alcohol 15ml, dipping spent the night 12-14 hour, and sonicated 25 minutes is put cold, supply the weight that subtracts mistake with methyl alcohol, shake up, filter, measure filtrate 5ml, steam near and do, adding neutral alumina 0.5g and mix thoroughly, add and put on the neutral alumina post, is 70% ethanol 80ml wash-out with mass concentration, collect eluent, be concentrated into driedly, the residue mass concentration is 50% dissolve with methanol, is transferred in the 5ml volumetric flask, and be diluted to 5ml, shake up, filter, get solution; Draw each 10 μ l of reference substance solution and capsule of weeping forsythia sample solution respectively, inject liquid chromatograph and measure, calculate the percentage of forsythin in the capsule of weeping forsythia sample;

C, set up the forsythin Quantitative Analysis Model

The forsythin content data that cold-maceration is recorded combines with its NIR spectroscopic data, sets up Quantitative Analysis Model with partial least-square regression method;

D, verification of model

The NIR spectroscopic data of 26 parts of forecast set samples is imported this Quantitative Analysis Model, obtain predicted value, compare with the value that records with cold-maceration, with the ratio between NIR predicted value and the pharmacopeia method measurement result as the prediction recovery, getting average recovery rate is 104.72%, with the capsule of weeping forsythia medicinal material forsythin content of accurate its coverage of prediction.

4, the method for utilizing near-infrared spectrum technique to estimate Chinese medicine medicinal material overall quality according to claim 1 is characterized in that said Chinese medicine medicinal material is a Chinese yam, and its near-infrared spectrum technique is realized by following steps the quality overall evaluation of Chinese yam:

The collection of the NIR diffuse reflection spectrum of Chinese yam medicinal material:

Collect from 115 parts of Chinese yam medicinal material samples of different cultivars, the different place of production, different collecting seasons;

After 115 parts of Chinese yam medicinal material oven dry, pulverized 100 mesh sieves respectively, put into rotating cup, the medicinal powder of tiling 2～5cm thickness uses the integrating sphere diffuse reflection, with near infrared spectrometer scan spectroscopic data, get wherein 105 parts as the calibration set sample, this is forecast set for 10 increments;

1) sets up Chinese yam medicinal material polysaccharide Quantitative Analysis Model

The selection of A, wave band and pre-service

By the TQ data processing software spectrogram is carried out wavelength and select, range of wavelengths is 7513.8～4597.8cm ^-1

B, the measurement of the polysaccharide content of Chinese yam medicinal material

Got the coarse yam powder 20g of No. two sieves, put in the apparatus,Soxhlet's, add 90 ℃ of backflow 4h of 250ml sherwood oil, after the taking-up coarse yam powder volatilizes, the adding mass concentration is 80% ethanol 200ml, reflux twice at 90 ℃, each two hours, to remove small-molecule substances such as monose, polyphenol, compound sugar and glycoside; After coarse yam powder volatilized solvent, add the water of 200ml, 80 ℃ of water-baths are extracted twice, the each extraction 4 hours merges extract, is concentrated into 50ml with the rotary evaporator vacuum, after removing albumen with the sevage method, vacuum is concentrated into 20ml once more, adds 95% ethanol, makes the concentration of alcohol in the concentrate of 20ml reach 80%, standing over night 12-14 hour, filter, residue successively washs successively with absolute ethyl alcohol, ether, acetone, and is dry below 60 ℃;

Take by weighing Chinese yam polysaccharide 10mg, be dissolved in the 100ml volumetric flask, be diluted to 100ml, shake up, it is an amount of to draw above-mentioned solution, puts in the 25ml volumetric flask, and adding distil water promptly gets the Chinese yam sample solution to 25ml; Record polyoses content with toluene method;

C, set up Chinese yam medicinal material polysaccharide Quantitative Analysis Model

The polyoses content data that toluene method is recorded combine with its NIR spectroscopic data, set up Quantitative Analysis Model with offset minimum binary (PLS) homing method;

D, verification of model:

The NIR spectroscopic data of 22 parts of forecast set samples is imported this Quantitative Analysis Model, obtain predicted value, compare, estimate this Quantitative Analysis Model, carry out internal chiasma checking RMSECV=1.18, R with the calibration samples collection with the value that records with toluene method ²=0.93;

2), set up Chinese yam medicinal material allantoin Quantitative Analysis Model

The selection of A, wave band and pre-service:

Selecting range of wavelengths by the TQ data processing software is 7513.8～6094.4cm ^-1With 5461.8～4242.9cm ^-1

B, Chinese yam medicinal material allantoin Determination on content

Adopt high-efficient liquid phase technique to record allantoin content in the Chinese yam medicinal material, and select wherein 105 increments this as the calibration set sample, 10 increments this as the forecast set sample, got 100 mesh sieves, the Chinese yam fine powder 1g of dry 3h below 60 ℃ puts in the ground conical flask, and adding mass concentration is 50% ethanol 50ml, ultrasonic Extraction 30min, place and to mend after 18-25 ℃ of the room temperature heavyly, filter, get filtrate 25ml in evaporating dish, water-bath concentrates to the greatest extent and does, water is dissolved to 10ml, and it is centrifugal in right amount to get root of large-flowered skullcap sample solution, gets supernatant sample introduction 5 μ l respectively, press chromatographic condition high effective liquid chromatography for measuring peak area, with the content of allantoin in the external standard method calculation sample;

C, set up Chinese yam medicinal material allantoin Quantitative Analysis Model

The allantoin content data that high performance liquid chromatography is recorded combines with its NIR spectroscopic data, sets up Quantitative Analysis Model with offset minimum binary (PLS) homing method;

D, verification of model

The NIR spectroscopic data of 10 parts of forecast set samples is imported this Quantitative Analysis Model, obtain predicted value, compare with the value that records with high performance liquid chromatography, the related coefficient of allantoin measured value and actual value is 0.9598, RMSEP is 0.0329, the measured value of allantoin and the absolute error between the actual value are between ± 0.1%, and what actual value was relatively concentrated is distributed near the center line;

3), set up Chinese yam medicinal material water extraction Quantitative Analysis Model:

The selection of A, wave band and pre-service:

Selecting range of wavelengths by the TQ data processing software is 7513.8～4242.9cm ^-1

B, Chinese yam medicinal material water extraction Determination on content:

Adopt high-efficient liquid phase technique to record water extraction content in the Chinese yam medicinal material, and select wherein 105 increments this as the calibration set sample, 10 increments this as the forecast set sample, adopt hot dipping: get the Chinese yam sample 4g of 60 ℃ of dryings, put in the conical flask of 250ml, add water 100ml, close plug, leave standstill 1 hour after, connect reflux condensing tube, be heated to boiling, and keep little and boiled 1 hour, after the cooling, take off conical flask, close plug, water is supplied the weight that subtracts mistake, shakes up, and filters with dry filter, measure filtrate 25ml, put in the dry evaporating dish, behind evaporate to dryness in the water-bath, in 105 ℃ of dryings 3 hours, put in the exsiccator and cooled off 30 minutes, calculate the content of water-soluble extractives in the Chinese yam sample;

C, set up Chinese yam medicinal material water extraction Quantitative Analysis Model

The water extraction content data that cold-maceration is recorded combines with its NIR spectroscopic data, sets up Quantitative Analysis Model with offset minimum binary (PLS) homing method;

D, verification of model

The NIR spectroscopic data of 10 parts of forecast set samples is imported this Quantitative Analysis Model, obtain predicted value, compare, estimate this Quantitative Analysis Model, carry out internal chiasma checking RMSECV=1.23, R with the calibration samples collection with the value that records with cold-maceration ²=0.94, best number of principal components is 12, and the absolute error between near infrared spectroscopy measured value and the actual value is between ± 3%.