1. The method for identifying the specific brand of Maotai-flavor liquor by adopting nuclear magnetic resonance and partial least square method based on non-volatile substances is characterized by comprising the following steps:

step one, establishing a data set of the Maotai-flavor liquor of a specific brand

A1, preparing N different batches of sauce-flavor liquor samples of specific brands, volatilizing the liquor samples, adding phosphate buffer salt solution to dissolve, wherein the phosphate buffer salt solution contains 3- (trimethylsilyl) deuterated sodium propionate as a calibration substance, centrifuging, and transferring the supernatant into a nuclear magnetic tube;

A2、¹h NMR detection: sample in nuclear magnetic tube¹H NMR detection, wherein detection data are recorded by 600M nuclear magnetic detection of a Bruk loading ultralow temperature probe, a standard one-dimensional composite pulse sequence is used for analyzing the wine sample, and pre-saturation is carried out during the relaxation delay period; the cycle delay time is 2s, the mixing time is 1s, the 90-degree pulse width of each sample is set to be 9.89 mu s, the frequency offset of a transmitter is 2820.78Hz, the pre-saturation power level is 23.00dB, the data point is 64k, the spectral width is 12019.230Hz, the scanning times are 128 times, the nuclear magnetic resonance signals of the Maotai-flavor liquor of a specific brand are obtained, and the non-volatile substances of different batches of specific Maotai-flavor liquor are established¹H NMR spectrum;

a3, spectrogram processing and derivation of measurement results: importing the acquired spectrogram into nuclear magnetic software MestReNova, flushing zero to 64K data points, correcting a baseline, carrying out integral range of 0.4-9.0 and integral interval of 0.002ppm, and carrying out normalization treatment on the total area to obtain a segmented integral result;

a4, establishing a specific brand Maotai-flavor liquor data set by taking the segmented integration result as an independent variable X matrix and different batches of specific brand Maotai-flavor liquor as a dependent variable Y matrix;

step two, establishing a data set of other brands of Maotai-flavor liquor

B1, preparing M sauce-flavor liquor samples of other brands, volatilizing the liquor samples, adding phosphate buffer salt solution to dissolve, wherein the phosphate buffer salt solution contains 3- (trimethylsilyl) deuterated sodium propionate as a calibration substance, centrifuging, and transferring supernatant into a nuclear magnetic tube;

b2, obtaining a sauce-flavor liquor data set of other brands according to the method of the steps A2-A4;

step three, constructing an identification model

C1, importing a working set obtained by integrating the specific brand Maotai-flavor liquor data set with other brand Maotai-flavor liquor data sets into SIMCA software, standardizing the data in the working set, dividing the data in the working set into t groups of specific brand Maotai-flavor liquor and f groups of other brand Maotai-flavor liquor, and then selecting and establishing a PLS-DA model to form an unmatched PLS-DA model;

c2, automatically fitting the PLS-DA model which is not fitted in the C1 in the SIMCA software, and selecting a component with the cumulative contribution rate of 99% to a variable as a main component to establish a partial least squares regression model according to a cross validity index to obtain an optimal PLS-DA model;

step four, evaluating and verifying PLS-DA model

Fraction R of variable Y explained by each component in the PLS-DA model²Y (cum) and the fraction Q of the predicted variable Y based on the cross-validated model²(cum) evaluating cumulative interpretability and cross-validation of the model;

the fitting degree of the fitted model is verified through an arrangement experiment, the arrangement frequency is 200 times, and p is<0.05；Q²Intercept at Y-axis<0.05, indicating that no overfitting of the model occurred; the correlation coefficient is within the range of 0.9-1, and the error between the predicted value and the actual value of the model is small;

step five, identifying unknown wine samples

Leading X unknown wine samples to be identified into the PLS-DA model established in the third step for prediction by a prediction set obtained by the method of A1-A4; and if the prediction result simultaneously meets the conditions that the prediction result is in a quadrant region where the Maotai-flavor liquor of the specific brand is located in the PLS-DA model, the goodness fit degree of the Maotai-flavor liquor of the specific brand is greater than 0.7 and is close to 1, determining that the unknown liquor sample is the Maotai-flavor liquor of the specific brand, and otherwise, determining that the unknown liquor sample is the Maotai-flavor liquor of the non-specific brand.

2. The method for identifying the specific brand of Maotai-flavor liquor based on the non-volatile substances by using nuclear magnetic resonance and partial least square method according to claim 1, which is characterized by comprising the following steps: and A1, drying the Maotai-flavor liquor sample by adopting nitrogen flow at the constant temperature of 37 ℃.

3. The method for identifying the specific brand of Maotai-flavor liquor based on the non-volatile substances by using nuclear magnetic resonance and the partial least square method according to claim 2, wherein the method comprises the following steps: in the step A1, the dosage of the wine sample is 5mL, the pH value of the phosphate buffer salt is 3, and the dosage is 600 uL.

4. The method for identifying the specific brand of Maotai-flavor liquor based on the non-volatile substances by using nuclear magnetic resonance and partial least square method according to claim 3, wherein the method comprises the following steps: in the step A1, the rotation speed is 14000rpm/min, and the time is 10 min.

Background

In recent years, the consumption of Maotai-flavor liquor (Maotai-flavor liquor for short) is very fierced nationwide, the Maotai-flavor liquor is burnt nationwide, and the Maotai-flavor liquor market becomes a main racetrack of the Maotai-flavor liquor market in China. The production process and the working procedures of the fake liquor can not meet the standard of the authentic Maotai-flavor liquor, the production process is unqualified, and the produced Maotai-flavor liquor has only slight difference in taste, but the harmfulness to the body is difficult to estimate. Meanwhile, for a specific brand of soy sauce, the brand image of the product is seriously damaged by a fake and real behavior in the market, and the healthy development of the soy sauce is also influenced.

Therefore, a set of scientific and effective true and false wine identification method is established, and the method has important significance for avoiding risks, reducing economic losses of enterprises and consumers, maintaining product brand images, guaranteeing safety and rights and interests of consumers, and standardizing the industry of Maotai-flavor liquor and the stable and healthy development of the market.

At present, the methods for identifying the Maotai-flavor liquor comprise a method based on volatile flavor substances, such as a gas chromatography method and an electronic nose, a method considering both volatile substances and nonvolatile substances, such as a method of a spectrum method, a liquid chromatography method, a mass spectrometry method, an electronic tongue and the like, and a method based on stable isotope ratio of the Maotai-flavor liquor, and the method is the most advanced in research by combining a multivariate statistical analysis method.

Although the non-volatile substances in the Maotai-flavor liquor do not generally present fragrance, the Maotai-flavor liquor has great influence on taste, mouthfeel, health functions and the like. However, the existing method for identifying the authenticity of the Maotai-flavor liquor has the following defects: (1) the research or the application of identifying the authenticity of the Maotai-flavor liquor aiming at the non-volatile substances of the Maotai-flavor liquor is not specially carried out. (2) The existing detection methods for the non-volatile substances of the Maotai-flavor liquor are all targeted detection of certain non-volatile substances, and the detection of the non-targeted non-volatile substances cannot be realized. (3) The existing method for detecting the non-volatile substances of the Maotai-flavor liquor generally needs derivatization treatment, which not only damages a sample and causes errors in an identification result, but also has complex operation in an experimental process. In summary, at present, no non-targeted Maotai-flavor liquor identification method based on Maotai-flavor liquor non-volatile substances exists.

Therefore, the method for detecting the non-volatile components in the Maotai-flavor liquor in a non-targeting manner is provided, and the method has great significance for identifying the authenticity of the Maotai-flavor liquor or whether an unknown liquor sample is the Maotai-flavor liquor of a specific brand.

Disclosure of Invention

Aiming at the defects in the prior art, the invention provides a method for identifying whether an unknown wine sample is a special brand of Maotai-flavor liquor by adopting a nuclear magnetic resonance hydrogen spectrum combined partial least square method based on non-volatile substances of the Maotai-flavor liquor.

The method for identifying the sauce-flavor liquor of the specific brand based on the non-volatile substances by adopting the nuclear magnetic resonance and partial least square method comprises the following steps:

step one, establishing a data set of the Maotai-flavor liquor of a specific brand

A1, preparing N different batches of sauce-flavor liquor samples of specific brands, volatilizing the liquor samples, adding phosphate buffer salt solution to dissolve, wherein the phosphate buffer salt solution contains 3- (trimethylsilyl) deuterated sodium propionate as a calibration substance, centrifuging, and transferring the supernatant into a nuclear magnetic tube;

A2、¹h NMR detection: sample in nuclear magnetic tube¹H NMR detection, wherein detection data are recorded by 600M nuclear magnetic detection of a Bruk loaded ultra-low temperature probe, a standard one-dimensional composite pulse sequence is used for analyzing the wine sample, and pre-saturation is carried out during relaxation delay; the cycle delay time is 2s, the mixing time is 1s, the 90-degree pulse width of each sample is set to be 9.89 mu s, the frequency offset of a transmitter is 2820.78Hz, the pre-saturation power level is 23.00dB, the data point is 64k, the spectral width is 12019.230Hz, the scanning times are 128 times, the nuclear magnetic resonance signals of the sauce-flavor liquor of a specific brand are obtained, and different brands are establishedNonvolatile substances of batch specific Maotai-flavor liquor¹H NMR spectrum;

a3, spectrogram processing and derivation of measurement results: importing the acquired spectrogram into nuclear magnetic software MestReNova, flushing zero to 64K data points, correcting a baseline, carrying out integral range of 0.4-9.0 and integral interval of 0.002ppm, and carrying out normalization treatment on the total area to obtain a segmented integral result;

a4, establishing a specific brand Maotai-flavor liquor data set by taking the segmented integration result as an independent variable X matrix and different batches of specific brand Maotai-flavor liquor as a dependent variable Y matrix;

step two, establishing a data set of other brands of Maotai-flavor liquor

B1, preparing M sauce-flavor liquor samples of other brands, volatilizing the liquor samples, adding phosphate buffer salt solution to dissolve, wherein the phosphate buffer salt solution contains 3- (trimethylsilyl) deuterated sodium propionate as a calibration substance, centrifuging, and transferring the supernatant into a nuclear magnetic tube;

b2, obtaining a sauce-flavor liquor data set of other brands according to the method of the steps A2-A4;

step three, constructing an identification model

C1, importing a working set obtained by integrating the specific brand Maotai-flavor liquor data set with other brand Maotai-flavor liquor data sets into SIMCA software, standardizing the data in the working set, dividing the data in the working set into t groups of specific brand Maotai-flavor liquor and f groups of other brand Maotai-flavor liquor, and then selecting and establishing a PLS-DA model to form an unmatched PLS-DA model;

c2, automatically fitting the PLS-DA model which is not fitted in the C1 in the SIMCA software, and selecting components with the accumulated contribution rate of more than 99% to variables as main components to establish a partial least squares regression model according to cross validity indexes to obtain the optimal PLS-DA model;

step four, evaluating and verifying PLS-DA model

Fraction R of variable Y explained by each component in the PLS-DA model²Y (cum) and the fraction Q of the predictor variable Y based on the cross-validated model²(cum) evaluating cumulative interpretability and cross-validation of the model;

the fitting degree of the fitted model is verified through an arrangement experiment, the arrangement frequency is 200 times, and p is<0.05；Q²Intercept at Y-axis<0.05, indicating that no overfitting of the model occurred; the correlation coefficient is within the range of 0.9-1, and the error between the predicted value and the actual value of the model is small;

step five, identifying unknown wine samples

And importing X unknown wine samples to be identified into the PLS-DA model established in the third step for prediction according to the method of A1-A4, wherein if the prediction results simultaneously meet the conditions that the quadrant region of the specific brand Maotai-flavor liquor in the PLS-DA model is in, the goodness of fit with the specific brand Maotai-flavor liquor is greater than 0.7 and close to 1, the unknown wine samples are the specific brand Maotai-flavor liquor, and if not, the unknown wine samples are the non-specific brand Maotai-flavor liquor.

The beneficial technical effect of this scheme is:

according to the characteristics of influence of non-volatile compounds of Maotai-flavor liquor on taste, mouthfeel, health functions and the like, the Maotai-flavor liquor non-volatile matter-based sauce-flavor liquor is utilized for the first time¹The H NMR detection technology is combined with a partial least square analysis method, the purpose of identifying whether an unknown wine sample is the Maotai-flavor liquor of a specific brand is achieved, compared with the existing detection method for the non-volatile substances of the Maotai-flavor liquor, the method has the advantages that not only is the sample not required to be subjected to derivatization, but also the non-targeted detection of the non-volatile substances in the Maotai-flavor liquor can be achieved. The method is important in that the method can realize the non-targeted detection of the non-volatile substances of the Maotai-flavor liquor based on the metabonomics thought, belongs to the field of first application, is highly innovative, and has excellent application effect in the identification of Maotai-flavor liquor of a specific brand.

By using the method, the unknown wine sample is identified, whether the detected wine sample is the sauce-flavor liquor of the specific brand is judged highly truly, and the risk is avoided to the greatest extent so as to reduce the economic loss of the wine enterprises of the specific brand.

Further, the Maotai-flavor liquor sample is dried in A1 by adopting nitrogen flow at the constant temperature of 37 ℃. The method can solve the problems that trace components are covered or difficult to detect and the like, but the method can correspondingly influence the signal peak of trace substances in the wine and further influence the authenticity of a detection result; therefore, in terms of reflecting the authenticity of the sample, the direct mode is to remove ethanol and water in the wine sample, the ethanol content of the Maotai-flavor liquor is relatively high, the Maotai-flavor liquor is not suitable for vacuum freeze drying, and the mode of removing ethanol by evaporation and then freeze drying is too tedious and time-consuming. Therefore, the scheme selects a nitrogen blow-drying mode to concentrate the Maotai-flavor liquor, the nitrogen blow operation is simple, a plurality of samples can be processed simultaneously, and the detection time can be greatly shortened; although the concentration can damage volatile substances in the Maotai-flavor liquor, the non-volatile substances in the Maotai-flavor liquor cannot be influenced by the concentration of the sample, and the Maotai-flavor liquor is a true reflection of the original liquor sample.

Further, the amount of the wine sample used in step A1 was 5mL, and the pH of the phosphate buffer salt was 3 at 600 uL.

Further, in the step A1, the centrifugation is carried out at 14000rpm for 10 min.

Drawings

FIG. 1 is a flow chart of the method for identifying a specific brand of Maotai-flavor liquor based on non-volatile substances by using nuclear magnetic resonance and partial least square method according to the invention;

FIG. 2 shows different batches of Maotai-flavor liquor of specific brands¹An enlarged view of the H NMR spectrum;

FIG. 3 is a PLS-DA model score plot;

FIG. 4 shows a PLS-DA model arrangement experiment;

FIG. 5 is a graph of validation scores of an unknown wine sample PLS-DA model.

Detailed Description

The following is further detailed by way of specific embodiments:

1. instruments, reagents and samples

1.1 Instrument: nuclear magnetic resonance spectrometer (Bruker, 600MHz), SIMCA software (umetics), centrifuge (edbend, germany), nuclear magnetic tube

1.2 reagent: phosphoric acid (greater than or equal to 85%, aladdin), sodium dihydrogen phosphate (greater than or equal to 99.0%, Macklin), sodium 3- (trimethylsilyl) deuterated propionate (98%, Sigma-Aldrich), and deuterium oxide (American CIL)

1.3 sample: and 33 finished products of Maotai-flavor liquor samples: the wine sample comprises 7 specific brands of Maotai-flavor white spirits (with the number of GB 1-GB 7), 15 other brands of Maotai-flavor white spirits (with the number of QT 1-QT 15), 11 unknown wine samples (with the number of YZ 1-YZ 11), and the design rules of the 11 unknown wine samples are as follows: one is the specific brand of Maotai-flavor liquor, and the others are all other brands of Maotai-flavor liquor.

2. Pretreatment of wine sample

Precisely absorbing 5mL of a sauce-flavor liquor sample, drying the liquor sample under a constant-temperature nitrogen flow at 37 ℃, adding 600uL of phosphate buffer salt (0.1M, pH 3) for dissolving, centrifuging the liquor for 10min at 14000rpm, wherein the phosphate buffer salt contains 3- (trimethylsilyl) deuterated sodium propionate as a calibration substance, and taking a supernatant for later use;

3. establishing a data set of the Maotai-flavor liquor of a specific brand

A1, and respectively taking 500uL of the supernatant of the 7 sauced wines with the specific brands, transferring the supernatant into a 5mm nuclear magnetic tube for detection;

A2、¹h NMR detection: sample data were recorded by 600M nuclear magnetic detection with a brook-loaded ultra-low temperature probe, the sample was analyzed using a standard one-dimensional complex pulse sequence (noesygppr1d) and pre-saturated during the relaxation delay; the cycle delay time is 2s, the mixing time is 1s, the 90-degree pulse width of each sample is set to be 9.89 mu s, the frequency deviation of the transmitter is 2820.78Hz, the pre-saturation power level is 23.00dB, the data point is 64k, the spectral width is 12019.230Hz, the scanning times are 128 times, the nuclear magnetic resonance signal of the Maotai-flavor liquor of a specific brand is obtained, and the non-volatile substances of different batches of specific Maotai-flavor liquor are established¹H NMR spectrum, as shown in figure 2;

a3, spectrogram processing and derivation of measurement results: importing the acquired spectrogram into MestReNova, flushing zero to 64K data points, correcting a baseline, carrying out normalization processing on the total area, wherein the integration range is 0.4-9.0, the integration interval is 0.002ppm, and obtaining a segmented integration result;

a4, establishing a specific brand Maotai-flavor liquor data set by taking the segmented integration result as an independent variable X matrix and different batches of specific brand Maotai-flavor liquor as a dependent variable Y matrix;

4. establishing other brand Maotai-flavor liquor data sets

B1, and respectively taking 500uL of the supernatant of the 15 other brands of the sauce wine, transferring the supernatant into a 5mm nuclear magnetic tube for detection;

b2, obtaining a sauce-flavor liquor data set of other brands according to the method of the steps A2-A4;

5. construction of authentication models

C1, importing a working set obtained by integrating the specific brand Maotai-flavor liquor data set with other brand Maotai-flavor liquor data sets into SIMCA software, standardizing the data in the working set, dividing the data in the working set into t groups of specific brand Maotai-flavor liquor and f groups of other brand Maotai-flavor liquor, and then selecting and establishing a PLS-DA model to form an unmatched PLS-DA model;

and C2, automatically fitting an unmatched PLS-DA model in C1 in the SIMCA software, selecting components with the cumulative contribution rate of more than 99% to variables as main components according to the cross validity index by the system to establish a partial least squares regression model to obtain the best PLS-DA model, wherein as shown in figure 3, the specific brand Maotai-flavor liquor obtains good distinction on a PLS-DA score chart, the specific brand Maotai-flavor liquor is concentrated in a third quadrant, other Maotai-flavor liquors are concentrated in the first, second and fourth quadrants, and the clustering effect is good.

6. Evaluation and validation of PLS-DA model

Fraction R of variable Y explained by each component in the PLS-DA model²Y (cum) and the fraction Q of the predictor variable Y based on the cross-validated model²(cum), the cumulative interpretability and cross-validation of the evaluation model, the results are shown in table 1; the fitting degree of the fitted model is verified through a permutation experiment, the permutation times are 200 times, p is less than 0.05, the PLS-DA model is verified as shown in figure 3, and the results are shown in table 2.

Q²The intercept at the Y axis is < 0.05, indicating that no overfitting of the model occurs; the closer the correlation coefficient is to 1, the smaller the error between the predicted value and the actual value of the model.

TABLE 1 PLS-DA model parameter information

Component	R2X	R2X(cum)	Eigenvalue	R2Y	R2Y(cum)	Q2	Limit	Q2(cum)	Significance	Iterations
											0	Cent.
1	0.135	0.135	2.97	0.689	0.689	0.386	0.05	0.386	R1	2
											2	0.0924	0.227	2.03	0.259	0.948	0.447	0.05	0.661	R1	2

TABLE 2 PLS-DA model arrangement experiment results List

As can be seen from tables 1 and 2, the correlation coefficient R of the PLS-DA model after fitting²Y (cum) is 0.948, Q²(cum) is 0.661, which shows that the PLS-DA model has better interpretability and cross-validation, the sample authenticity and the interpretation variable have obvious linear relation, and the model verification is shown in FIG. 4, Q²Has a regression line intercept of-0.22, less than 0.05, and all left Q²The values are all lower than the original points on the rightmost side, and no overfitting phenomenon exists, so that the reasonability and the accuracy of the model are proved.

7. Differential prediction of unknown wine samples

A data set formed by 11 unknown wine samples according to the methods of A1-A4 is imported into an established PLS-DA model for prediction, and the prediction results are shown in Table 3 and FIG. 5.

As can be seen from fig. 5 and table 3, only YZ7 of the 11 verification samples is located in the third quadrant, and the degree of coincidence with the sauce-flavor liquor of a specific brand is 1.08, which is close to 1.0, i.e., in the range of 0.7-1.3. The other verification samples except YZ7 are all positioned in the first quadrant, the second quadrant and the fourth quadrant, and the matching degree with the sauce-flavor white spirit of a specific brand is less than 0.7. The result shows that YZ7 is the Maotai-flavor liquor of a specific brand, and other liquor samples are Maotai-flavor liquor of unspecific brands; the total shows that 11 unknown wine samples (1 specific brand and 10 other brands) are correctly classified, and the total authentication accuracy is 100%.

TABLE 3 validation results of the PLS-DA model for unknown wine samples

Sample numbering	Degree of match with the sauce wine (t) of a specific brand	Goodness of fit with other brands of soy sauce wine (f)	Verification result
				YZ1	0.009169	0.990831	f
YZ2	0.244550	0.755450	f
				YZ3	0.196000	0.804000	f
YZ4	-0.237833	1.237830	f
				YZ5	-0.097571	1.097570	f
YZ6	-0.112115	1.112120	f
				YZ7	1.082600	-0.082604	t
YZ8	0.098670	0.901330	f
				YZ9	-0.073214	1.073210	f
YZ10	0.158764	0.841236	f
				YZ11	0.099371	0.900629	f