CN104237157A - Method for utilizing terahertz time-domain spectroscopy technology to detect amino acid content in grain - Google Patents

Abstract

The invention provides a method for utilizing a terahertz time-domain spectroscopy technology to detect amino acid content in grain. The method comprises the following steps: grinding and then performing tableting on a grain sample to be detected, directly utilizing a terahertz time-domain spectroscopy system to detect the sample subject to tableting in a transmission measurement mode and under the nitrogen condition to obtain terahertz time-domain spectroscopy signals as sample signals, under the same conditions, acquiring the terahertz time-domain spectroscopy signals under the nitrogen condition as reference signals, then obtaining an absorptivity spectrum of the grain tableting sample to be detected, at last dividing the absorptivity spectrum of the grain tableting sample to be detected into a calibration set sample absorptivity spectrum and a verification set sample absorptivity spectrum, utilizing a partial least squares regression method to establish a quantitative analysis mode, and obtaining the quantitative detection value of the grain sample to be detected. The method disclosed by the invention is simple in preparation of the sample, is not required to process any pretreatment, and can practically and effectively realize fast and accurate quantitative detection on the amino acid in the grain or food.

Description

Application terahertz time-domain spectroscopic technology detects the method for Amino-Acid in Grain content

Technical field

The present invention relates to a kind of method applying terahertz time-domain spectroscopic technology detection Amino-Acid in Grain content, belong to food composition detection technique field.

Background technology

Amino acid is the important nutrient of growth in humans, is the basic composition unit of polypeptide and biological function macro-molecular protein in constituting body.Amino acid not only has unique physiological function, also has unique function in the food industry simultaneously.Apply more amino acid in the food industry and have glutamic acid, lysine, threonine, tryptophane, arginine etc.Wherein, Pidolidone is due to the flavor of its uniqueness, and mainly for the production of monosodium glutamate, spices etc., and Pidolidone is also important nutritional supplement and biochemical reagents, itself as medicine, can participate in the metabolism of brain internal protein and sugar, accelerating oxidation process.Therefore, need to find and a kind of detect amino acid whose detection method in food fast and accurately, thus provide foundation for the evaluation of amino acid content in food.

Chinese patent literature CN102590129A discloses a kind of method detecting amino acid content in peanut, concrete steps are as follows: (1) Standard for Peanuts product to known amino acid content carry out near infrared spectrum scanning, obtain all spectral informations of Standard for Peanuts product at near-infrared wavelength of described known amino acid content, obtain the calculating mean value of calibration set sample spectrum; (2) pre-service is carried out to described step (1) gained calibration set sample spectrum; (3) principal component analysis (PCA) is carried out, characteristic information extraction data to the information data in described step (2) pretreated calibration set sample spectrum; (4) with the chemical measurements of the amino acid content of described Standard for Peanuts product for corrected value, using described step (3) gained characteristic information data as independent variable, described corrected value, as dependent variable, sets up the calibration model between described independent variable and described dependent variable with Chemical Measurement Multivariate Correction algorithm; (5) the Standard for Peanuts product of described step (1) described known amino acid content are replaced with peanut sample to be measured, repeating said steps (1)-step (3), described step (3) gained characteristic information data is inputted the calibration model of described step (4), obtain the amino acid content in described peanut sample to be measured.But above-mentioned employing near infrared spectrum detects the method for amino acid content in peanut, and its Problems existing is: near infrared spectrum be absorbs with sum of fundamental frequencies based on the frequency multiplication of intramolecule vibration, spectrum peak is wider, and overlap is seriously and absorption intensity is weak.Because the spectrum peak of near infrared spectrum is wider, directly can not recognize the individual features absorption peak of material near infrared spectrum middle finger, thus need to utilize full modal data Modling model, and then easily introduce more noise information.The penetrability of near infrared spectrum is relatively weak simultaneously, cannot penetrate and there is certain thickness coating or external packing, for when detecting below coating or there is the sample of external packing, need to destroy coating or external packing, thus destroy the integrality of sample to a certain extent.

Terahertz emission (also claiming " THz radiation ") refers to that frequency is at 0.1THz-10THz, the electromagnetic wave of wavelength between 0.03-3mm, its wave band is between microwave and infrared ray, be the region of macroelectronics to the transition of microcosmic photonics, in electromagnetic spectrum, occupy very special position.The transition of the large molecule of much polarity between vibrational energy level is just in time in Terahertz frequency range, and therefore, the tera-hertz spectra of biomolecule can reflect by molecule or the intrinsic property of low frequency diaphragm that causes of intermolecular collective vibration and lattice vibration.Terahertz electromagnetic wave has lower photon energy, when carrying out sample detection, can not produce harmful photoionization.In addition, THz wave has stronger penetrability, and can penetrate certain thickness interlayer and detect analysis sample, can avoid destroying precious sample, as artifact etc., be a kind of novel effective lossless detection method.But, also terahertz light spectral technology is not used for the relevant report that amino acid detects at present.

Summary of the invention

Technical matters to be solved by this invention is that providing a kind of applies the method that terahertz time-domain spectroscopic technology realizes directly carrying out Amino-Acid in Grain quantitatively detection.

For solving the problems of the technologies described above, the present invention is achieved by the following technical solutions:

Utilize terahertz time-domain spectroscopic technology to detect a method for Amino-Acid in Grain content, comprise the steps:

(1) get the rear compressing tablet of grain samples to be measured grinding, obtain grain to be measured or food press sheet compression;

(2) utilize terahertz time-domain spectroscopy system, the terahertz time-domain spectroscopy signal under its nitrogen atmosphere of employing transmission measurement type collection is as reference signal E _ref(t), and the terahertz time-domain spectroscopy signal gathering described grain press sheet compression to be measured is under the same conditions as sample signal E _sam(t), Reference Signal E _ref(t) and sample signal E _samt () is respectively through the frequency-region signal E obtaining reference after Fourier transform _ref(ω) and the frequency-region signal E of sample _sam(ω), with reference to frequency-region signal E _ref(ω) and the frequency-region signal E of sample _sam(ω) ratio is defined as function H (ω), see formula (1), based on function H (ω), calculate refractive index n (ω) and absorption coefficient spectrum α (ω) of sample, see formula (2) and formula (3), and according to the interval characteristic wave bands as described target amino acid of the good wave band of reappearance that the character of target amino acid selects described absorption coefficient to compose:

$\mathrm{\α}\left(\mathrm{\ω}\right)=\frac{2\mathrm{\ωk}\left(\mathrm{\ω}\right)}{c}=\frac{2}{d}\ln \left[\frac{4n\left(\mathrm{\ω}\right)}{\mathrm{\ρ}\left(\mathrm{\ω}\right){(n\left(\mathrm{\ω}\right)+1)}^2}\right]---\left(3\right)$

In above-mentioned formula (1)-(3), n ＇ is the complex index of refraction of grain press sheet compression to be measured, and k (ω) is the extinction coefficient under each frequency, d is the thickness of grain press sheet compression to be measured, ω is frequency, and ρ (ω) is amplitude represent phase place, c represents the velocity of propagation of light in vacuum;

(3) in selected described characteristic wave bands, the absorption coefficient of described grain press sheet compression to be measured spectrum is divided into calibration set sample absorbance coefficient spectrum and checking collection sample absorbance coefficient spectrum;

(4) utilize partial least-squares regression method to set up the Quantitative Analysis Model of described calibration set sample absorbance coefficient spectrum and described checking collection sample absorbance coefficient spectrum, obtain amino acid whose quantitative detected value in each described grain samples to be measured.

Described amino acid is Pidolidone.

Described grain samples is rice.

In described step (2), described characteristic wave bands is 0.3-1.8THz.

In described step (2), the test condition of described terahertz time-domain spectroscopy is: temperature is 20 DEG C, and frequency range is 0.3-3.0THz.

In described step (3), the division of described calibration set sample and described checking collection sample absorbance coefficient spectrum adopts Latin partition method to carry out.

Select partition number to be 4 when utilizing described Latin partition method to divide, get wherein 3/4 as calibration set sample, 1/4 as checking collection sample.

The circle sheet that described grain press sheet compression to be measured is diameter 13mm, thickness is about 1.0mm.

In described step (4), adopt the main cause subnumber of cross-validation method determination partial least squares regression.

In described step (4), adopt square error MSE and square coefficient R of test set ²evaluate the estimated performance accuracy of institute's Modling model, described MSE and R ²computing formula as (4) and (5):

$\mathrm{MSE}=\frac{1}{n}{\mathrm{\Σ}}_{i=1}^n{(f\left(x_i\right)-y_i)}^2---\left(4\right)$

$R^2=\frac{{(n{\mathrm{\Σ}}_{i=1}^{n}f\left(x_i\right)y_i-{\mathrm{\Σ}}_{i=1}^{n}f\left(x_i\right){\mathrm{\Σ}}_{i=1}^n{y}_i)}^2}{(n{\mathrm{\Σ}}_{i=1}^{n}f{\left(x_i\right)}^2-{\left({\mathrm{\Σ}}_{i=1}^{n}f\left(x_i\right)\right)}^2)(n{\mathrm{\Σ}}_{i=1}^n{y}_i^2-{\left({\mathrm{\Σ}}_{i=1}^n{y}_i\right)}^2)}---\left(5\right)$

Technique scheme of the present invention has the following advantages compared to existing technology:

(1) a kind of method applying amino acid content in terahertz time-domain spectroscopic technology detection grain and food of the present invention, by the grain press sheet compression to be measured that compressing tablet after grain samples grinding to be measured is obtained, direct employing terahertz time-domain spectroscopy system is in transmission measurement pattern, under nitrogen atmosphere, it is tested, obtain terahertz time-domain spectroscopy signal as sample signal, terahertz time-domain spectroscopy signal under nitrogen atmosphere is gathered as with reference to signal under similarity condition, and then obtain the absorption coefficient spectrum of described grain press sheet compression to be measured, obtain the characteristic wave bands of described target amino acid, finally the absorption coefficient of described grain press sheet compression to be measured spectrum is divided into calibration set sample absorbance coefficient spectrum and checking collection sample absorbance coefficient spectrum, Quantitative Analysis Model is set up by partial least squares regression, obtain the quantitative detected value of each described grain samples to be measured.The method of the invention does not need to mix other any materials in the sample that detects, sample preparation is simple, do not need to carry out any pre-service, can truly, effectively realize quantitatively detecting fast and accurately the amino acid in grain or food, the prediction square error of described Quantitative Analysis Model is less, squared correlation coefficient (R ²) up to 0.9956.

(2) a kind of method applying amino acid content in terahertz time-domain spectroscopic technology detection grain and food of the present invention, before employing partial least square method sets up calibration model, Latin partition method is adopted to be divided into calibration set and checking collection to described absorption coefficient spectrum, can accurate evaluation quantitative calibration models predictive ability and stability; Latin partition method described here is a kind of model performance verification method be based upon on cross validation and random sampling checking basis.Latin partition method can realize uniform random sampling checking, every partition once, each sample as and only as one-time authentication collection; When partition number is 4, each sample be used for and only for one-time authentication collection, and three times as calibration set, set up four models respectively, ensure that each sample is concentrated at calibration set and checking to occur with same ratio, thus realize evaluating without inclined institute's established model predictive ability, make quantitative calibration models more reliable, analysis result has more statistical significance.

(3) a kind of method applying amino acid content in terahertz time-domain spectroscopic technology detection grain and food of the present invention, in described step (4), to described calibration set sample absorbance coefficient spectrum and described checking collection sample absorbance coefficient spectrum, and corresponding concentration value, partial least squares regression is adopted to set up Quantitative Analysis Model, and adopt leaving-one method as the main cause subnumber of cross validation method determination partial least squares regression, thus sample nearly all in every bout is all for training pattern, the most close with the distribution of original sample, analysis result accuracy is high, owing to there is not any enchancement factor affecting experimental data in experimentation, experimentation can be replicated, and under the condition of best main cause subnumber, set up Partial Least-Squares Regression Model, farthest can retain the useful information in original spectral data, introduce measurement noises as few as possible simultaneously.

Accompanying drawing explanation

In order to make content of the present invention be more likely to be clearly understood, below according to a particular embodiment of the invention and by reference to the accompanying drawings, the present invention is further detailed explanation, wherein

Fig. 1 is that the inventive method measures Pidolidone-poly Terahertz absorption coefficient spectrum;

Fig. 2 is the absorption coefficient spectrum of part rice sample to be measured in 0.3-1.8THz characteristic wave bands interval described in the embodiment of the present invention 1;

Fig. 3 is the root-mean-square error of leave one cross validation described in the inventive method and the graph of a relation of PLS main cause subnumber;

Fig. 4 is the graph of a relation between the model predication value of Pidolidone in rice of the present invention and experiment value.

Embodiment

Embodiment

The present embodiment provides a kind of method utilizing terahertz time-domain spectroscopic technology to analyze Pidolidone content in grain or food.Wherein, the biased sample to be measured of Different L-aminoglutaric acid concentration is prepared after rice powder (Zhangjiakou City inspection and quarantine bureau of Hebei province provides) interpolation known quantity Pidolidone (being purchased from Sigma company), and adopt the inventive method to carry out detection analysis as blind sample described testing sample, concrete steps are as follows:

(1) described rice is put into comminutor to pulverize, sieve after grinding (100 orders, particle diameter≤150 μm), put into baking oven and dry, obtain rice powder, rice powder is mixed according to different quality ratio with Pidolidone powder, be transferred in agate mortar to grind further and obtain described biased sample fine powder to be measured, wherein, in described biased sample to be measured, Pidolidone mass percentage is followed successively by 0.00%, 0.50%, 1.00%, 1.50%, 2.00%, 2.50%, 2.96%, 3.50%, 4.00%, 4.27%, 5.00%, 5.50%, 6.00%, 6.45%, 7.01%, 7.50%, 8.01%, 8.50%, 9.02%, 9.51%, 10.00%, 11.01%, 12.02%, 13.07%, 14.04%, 14.88%, 16.00%, 17.02%, 18.00%, 19.02%, 20.04%, amount to 31 groups,

Take above-mentioned often kind of described biased sample fine powder to be measured and be about 170mg, be placed in the mould of Specac company, under the pressure of 5t, 3-4min is kept with sheeter, thus by the described circle sheet that biased sample fine powder to be measured is pressed into diameter 13mm, thickness is about 1.0mm, obtain the grain press sheet compression to be measured containing different quality number percent Pidolidone, described grain press sheet compression two to be measured surface be parallel, smooth surface and do not have crack.

(2) the terahertz time-domain spectroscopy system of standard is utilized, T-Spectroscopy signals collecting software, under employing U.S. Newport house flag, the Mai Tai titanium sapphire femtosecond laser oscillator of Spectra-Physics brand is as lasing light emitter, pulse center wavelength 800nm, output power > 500mW, under 20 DEG C of conditions, Terahertz frequency range is 0.3-3.0THz, with nitrogen as a reference, adopt transmission measurement pattern, gather the terahertz time-domain spectroscopy signal of described grain press sheet compression to be measured as sample signal E _sam(t), and before the time-domain spectroscopy data in Terahertz region gathering each press sheet compression, first gather the time-domain spectroscopy signal in the Terahertz region under the nitrogen atmosphere not placing press sheet compression as reference signal E _ref(t); Reference Signal E _ref(t) and sample signal E _samt () is respectively through the frequency-region signal E obtaining reference after Fast Fourier Transform (FFT) _ref(ω) and the frequency-region signal E of sample _sam(ω), with reference to frequency-region signal E _ref(ω) and the frequency-region signal E of sample _sam(ω) ratio is defined as function H (ω), see formula (1), based on function H (ω), refractive index n (ω) and absorption coefficient spectrum α (ω) of sample can be extrapolated further, see formula (2) and formula (3):

$\mathrm{\α}\left(\mathrm{\ω}\right)=\frac{2\mathrm{\ωk}\left(\mathrm{\ω}\right)}{c}=\frac{2}{d}\ln \left[\frac{4n\left(\mathrm{\ω}\right)}{\mathrm{\ρ}\left(\mathrm{\ω}\right){(n\left(\mathrm{\ω}\right)+1)}^2}\right]---\left(3\right)$

In formula, n ＇ is the complex index of refraction of grain press sheet compression to be measured, and k (ω) is the extinction coefficient under each frequency, and d is the thickness of grain press sheet compression to be measured, and ω is frequency, and ρ (ω) is amplitude, represent phase place, c represents the velocity of propagation of light in vacuum.

When carrying out above-mentioned data acquisition, the sample spectra of each described grain press sheet compression to be measured repeated acquisition 3 times under equivalent environment, the mean value finally getting 3 times, as subsequent treatment absorption coefficient modal data used, finally obtains 31 groups of absorption coefficient modal data.

Be illustrated in figure 1 and adopt the inventive method to measure Pidolidone-poly Terahertz absorption coefficient spectrum, can find out, in the frequency range of 0.3-1.8THz, there is obvious characteristic absorption being positioned at frequency 1.23THz place in Pidolidone, therefore, the interval 0.3-1.8THz of the good wave band of the reappearance selecting described absorption coefficient to compose is as the characteristic wave bands of described target agricultural chemicals.

Be illustrated in figure 2 the absorption coefficient spectrum of grain samples described to be measured in 0.3-1.8THz characteristic wave bands interval of wherein 10 Different L-glutamic acid mass percentage, 10 Pidolidone mass percentage are respectively 0.00%, 1.00%, 2.96%, 5.00%, 7.01%, 10.00%, 13.08%, 14.88%, 17.02%, 20.04%; In addition, as can be seen from Figure 2, in the scope that Pidolidone massfraction is 0%-20%, the characteristic absorption coefficient of Pidolidone and the change in direct ratio of its massfraction, thus the Terahertz characteristic absorption coefficient spectrum detecting Pidolidone can be used in the quantitative test of Pidolidone in grain.

(3) in the characteristic wave bands of selected grain press sheet compression described to be measured, the absorption coefficient of described grain press sheet compression to be measured is composed, above-mentioned 31 groups of absorption coefficient modal data, self-service Latin partition method is adopted to be divided into calibration set sample absorbance coefficient spectrum and test set sample absorbance coefficient spectrum, partition number is selected to be 4, get wherein 3/4 as calibration set sample, 1/4 as test set sample, be specially: first sample to be tested is divided into 4 parts, select wherein 1 part as test set sample, all the other 3 parts as calibration set sample, it should be noted that, in each calculating, each sample is only for once predicting checking.

(4) calibration set data set is utilized, quantitative calibration models is set up in conjunction with partial least squares regression, adopt leave one cross validation, determine the main cause subnumber of partial least squares regression, as shown in Figure 3, when main gene is 4, the root-mean-square error of cross validation is minimum, RMSECV=0.0036 (MSECV=1.296 × 10 ^-5), adopt 4 main genes when therefore setting up quantitative calibration models.The concrete principle of described partial least squares regression is as follows:

First partial least squares regression decomposes the Terahertz absorption coefficient spectrum matrix X of sample and concentration matrix Y, and its model tormulation is as follows:

$Y={\mathrm{UQ}}^T+E_Y=\underset{i=1}{\overset{k}{\mathrm{\Σ}}}{u}_k{q}_K^T+E_Y$

$X={\mathrm{TP}}^T+E_X=\underset{i=1}{\overset{k}{\mathrm{\Σ}}}{t}_k{p}_k^T+E_X$

In above-mentioned expression formula, t _k(n × 1) is the score of i-th main gene of absorption coefficient matrix X; p _k(1 × m) is the load of i-th main gene of absorption coefficient matrix X; u _k(n × 1) is the score of i-th main gene of concentration matrix Y, q _k(1 × p) is the load of i-th main gene of concentration matrix Y; K is main cause subnumber.T and U is the score matrix of X and Y matrix respectively, P and Q is the loading matrix of X and Y matrix respectively, Ex and E _ythen the PLS regression criterion matrix of X and Y matrix respectively.

Afterwards T and U two score matrix T and U are done linear regression:

U＝TB

B＝(T ^TT) ^-1T ^TY

Last when predicting, first obtain the score matrix T ＇ of unknown sample absorption coefficient matrix X ＇ according to the loading matrix P of absorption coefficient matrix X, then can be obtained the concentration prediction matrix Y ＇ of unknown sample by following formula:

Y＇＝T＇BQ

The concentration prediction value y ＇ of the unknown sample in concentration prediction matrix Y ＇ can be obtained thus;

(5) test set absorption coefficient spectrum data set is utilized, under best main cause subnumber, verify the estimated performance of the partially minimum second metering calibration model set up, the actual concentrations of different sample and the comparing result of prediction concentrations as shown in table 1, result shows, within the scope of the Pidolidone massfraction of 0-20%, the prediction square error of partial least square model is MSE=1.444 × 10 ^-5, squared correlation coefficient R ²=0.9956.Described MSE and R ²computing formula as follows:

$\mathrm{MSE}={\frac{1}{n}\underset{i=1}{\overset{n}{\mathrm{\Σ}}}(f\left(x_i\right)-y_i)}^2$

$R^2=\frac{{(n{\mathrm{\Σ}}_{i=1}^{n}f\left(x_i\right)y_i-{\mathrm{\Σ}}_{i=1}^{n}f\left(x_i\right){\mathrm{\Σ}}_{i=1}^n{y}_i)}^2}{(n{\mathrm{\Σ}}_{i=1}^{n}f{\left(x_i\right)}^2-{\left({\mathrm{\Σ}}_{i=1}^{n}f\left(x_i\right)\right)}^2)(n{\mathrm{\Σ}}_{i=1}^n{y}_i^2-{\left({\mathrm{\Σ}}_{i=1}^n{y}_i\right)}^2)}.$

Be illustrated in figure 4 described model about the graph of a relation between the predicted value of Pidolidone and experiment value, can know and find out that the correlativity between described model predication value and experiment value is satisfactory.

The different sample actual concentrations value of table 1-and prediction concentrations value

As can be seen here, the MSE of above-mentioned analytical model prediction is less, R ²up to 0.9956, thus illustrate that the model that the inventive method is set up is reliably feasible, can be used for quantitatively detecting Pidolidone in rice sample.

Adopt the method for foregoing description, not only the Pidolidone in rice quantitatively can be detected, also can be generalized to the amino acid whose detection of other kind.In the detection of the sample of other kind grain, also need sample preparation powdered compressing tablet.As long as and the target amino acid being suitable for detecting has characteristic absorption in Terahertz frequency range and available the method for the invention realizes detecting.Detection method of the present invention can be accurate, easy the aminoacid ingredient in grain is detected, obtained the subsidy of the great scientific instrument special project (2012YQ140005) of state natural sciences fund (21275101) and country.

Obviously, above-described embodiment is only for clearly example being described, and the restriction not to embodiment.For those of ordinary skill in the field, can also make other changes in different forms on the basis of the above description.Here exhaustive without the need to also giving all embodiments.And thus the apparent change of extending out or variation be still among the protection domain of the invention.

Claims (10)

1. utilize terahertz time-domain spectroscopic technology to detect a method for Amino-Acid in Grain content, it is characterized in that, comprise the steps:

(1) get the rear compressing tablet of grain samples to be measured grinding, obtain grain to be measured or food press sheet compression;

(2) utilize terahertz time-domain spectroscopy system, the terahertz time-domain spectroscopy signal under its nitrogen atmosphere of employing transmission measurement type collection is as reference signal E _ref(t), and the terahertz time-domain spectroscopy signal gathering described grain press sheet compression to be measured is under the same conditions as sample signal E _sam(t), Reference Signal E _ref(t) and sample signal E _samt () is respectively through the frequency-region signal E obtaining reference after Fourier transform _ref(ω) and the frequency-region signal E of sample _sam(ω), with reference to frequency-region signal E _ref(ω) and the frequency-region signal E of sample _sam(ω) ratio is defined as function H (ω), based on function H (ω), calculate refractive index n (ω) and absorption coefficient spectrum α (ω) of sample, and according to the interval characteristic wave bands as described target amino acid of the good wave band of reappearance that the character of target amino acid selects described absorption coefficient to compose:

$\mathrm{\α}\left(\mathrm{\ω}\right)=\frac{2\mathrm{\ωk}\left(\mathrm{\ω}\right)}{c}=\frac{2}{d}\ln \left[\frac{4n\left(\mathrm{\ω}\right)}{\mathrm{\ρ}\left(\mathrm{\ω}\right){(n\left(\mathrm{\ω}\right)+1)}^2}\right]---\left(3\right)$

In above-mentioned formula (1)-(3), n ＇ is the complex index of refraction of grain press sheet compression to be measured, and k (ω) is the extinction coefficient under each frequency, d is the thickness of grain press sheet compression to be measured, ω is frequency, and ρ (ω) is amplitude represent phase place, c represents the velocity of propagation of light in vacuum;

(3) in selected described characteristic wave bands, the absorption coefficient of described grain press sheet compression to be measured spectrum is divided into calibration set sample absorbance coefficient spectrum and checking collection sample absorbance coefficient spectrum;

(4) utilize partial least-squares regression method to set up the Quantitative Analysis Model of described calibration set sample absorbance coefficient spectrum and described checking collection sample absorbance coefficient spectrum, obtain amino acid whose quantitative detected value in each described grain samples to be measured.

2. a kind of method utilizing terahertz time-domain spectroscopic technology to detect amino acid content in grain and food according to claim 1, it is characterized in that, described amino acid is Pidolidone.

3. a kind of method applying amino acid content in terahertz time-domain spectroscopic technology detection grain and food according to claim 2, it is characterized in that, described grain samples is rice.

4. a kind of method applying amino acid content in terahertz time-domain spectroscopic technology detection grain and food according to claim 3, it is characterized in that, in described step (2), described characteristic wave bands is 0.3-1.8THz.

5. according to the arbitrary a kind of described method applying amino acid content in terahertz time-domain spectroscopic technology detection grain and food of claim 1-4, it is characterized in that, in described step (2), the test condition of described terahertz time-domain spectroscopy is: temperature is 20 DEG C, and frequency range is 0.3-3.0THz.

6. according to the arbitrary a kind of described method applying amino acid content in terahertz time-domain spectroscopic technology detection grain and food of claim 1-5, it is characterized in that, in described step (3), the division of described calibration set sample and described checking collection sample absorbance coefficient spectrum adopts Latin partition method to carry out.

7. according to the arbitrary a kind of described method applying amino acid content in terahertz time-domain spectroscopic technology detection grain and food of claim 1-6, it is characterized in that, partition number is selected to be 4 when utilizing described Latin partition method to divide, get wherein 3/4 as calibration set sample, 1/4 as checking collection sample.

8., according to the arbitrary a kind of described method applying amino acid content in terahertz time-domain spectroscopic technology detection grain and food of claim 1-7, it is characterized in that, the circle sheet that described grain press sheet compression to be measured is diameter 13mm, thickness is about 1.0mm.

9. according to the arbitrary a kind of described method applying amino acid content in terahertz time-domain spectroscopic technology detection grain and food of claim 1-8, it is characterized in that, in described step (4), adopt the main cause subnumber of cross-validation method determination partial least squares regression.

10., according to the arbitrary a kind of described method applying amino acid content in terahertz time-domain spectroscopic technology detection grain and food of claim 1-9, it is characterized in that, in described step (4), adopt square error MSE and square coefficient R of test set ²evaluate the estimated performance accuracy of institute's Modling model, described MSE and R ²computing formula as (4) and (5):

$\mathrm{MSE}=\frac{1}{n}{\mathrm{\Σ}}_{i=1}^n{(f\left(x_i\right)-y_i)}^2---\left(4\right)$

$R^2=\frac{{(n{\mathrm{\Σ}}_{i=1}^{n}f\left(x_i\right)y_i-{\mathrm{\Σ}}_{i=1}^{n}f\left(x_i\right){\mathrm{\Σ}}_{i=1}^n{y}_i)}^2}{(n{\mathrm{\Σ}}_{i=1}^{n}f{\left(x_i\right)}^2-{\left({\mathrm{\Σ}}_{i=1}^{n}f\left(x_i\right)\right)}^2)(n{\mathrm{\Σ}}_{i=1}^n{y}_i^2-{\left({\mathrm{\Σ}}_{i=1}^n{y}_i\right)}^2)}---\left(5\right)$