1. A method for predicting the adulterated oil type in tea oil is characterized by comprising the following steps:

(S1) sample preparation: mixing the adulterated oil and the adulterated oil according to different mass proportions by taking the tea oil as the adulterated oil and other edible oil as the adulterated oil to obtain an experimental sample of the adulterated tea oil, wherein the experimental sample is divided into a training set and a verification set;

(S2) data acquisition: collecting the gas chromatogram of the experimental sample as fatty acid basic data, and collecting the liquid chromatogram of the experimental sample as tocopherol basic data;

(S3) unsupervised model building: based on unsupervised HCA analysis, respectively combining fatty acid basic data and tocopherol basic data to establish a corresponding pure oil distinguishing visual model;

(S4) supervised model building: simultaneously taking fatty acid basic data and tocopherol basic data in a training set as input variables, and respectively combining with a supervised PLS-DA and/or SIMCA model to establish a corresponding PLS-DA discrimination model and/or SIMCA discrimination model;

(S5) model verification: and respectively verifying the PLS-DA discrimination model and/or the SIMCA discrimination model by using the fatty acid basic data and the tocopherol basic data which are centralized in verification, and verifying the accuracy of the PLS-DA discrimination model and/or the SIMCA discrimination model.

2. The method of claim 1, wherein the fatty acid basis data comprises fatty acid ratio and the tocopherol basis data comprises tocopherol composition.

3. The method of claim 2, wherein the fatty acid ratio values comprise P/S, O/P, O/L, O/Ln, and L/S; the tocopherol composition includes mass percentages of alpha-tocopherol, beta-tocopherol, gamma-tocopherol, and delta-tocopherol, and a ratio of gamma/alpha of gamma-tocopherol to alpha-tocopherol.

4. The method of claim 1, wherein the adulteration ratio of the test sample of adulterated tea oil in (S1) is 0%, 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 100% (w/w), and the adulterated oil comprises any one of corn oil, rapeseed oil, rice bran oil, sesame oil, and soybean oil.

5. The method according to claim 4, wherein the experimental sample of adulterated tea oil in (S1) is: optionally selecting 3 kinds of tea oil from 36 kinds of tea oil as sample tea oil, 3 kinds of corn oil, 3 kinds of rapeseed oil, 3 kinds of rice bran oil, 3 kinds of sesame oil and 3 kinds of soybean oil, and mixing with the three kinds of tea oil according to the proportion to obtain 135 adulterated samples, and 191 samples in total.

6. The method of claim 1, wherein the collecting of the gas chromatogram of the experimental sample as fatty acid-based data in (S2) comprises: extracting fatty acid of the experimental sample, carrying out gas chromatography on the fatty acid to obtain a corresponding fatty acid map, and acquiring basic data of the fatty acid;

the fatty acids from which the experimental sample is extracted include: carrying out methyl esterification on the experimental sample doped with the pseudo tea oil, and carrying out microfiltration on supernate to obtain fatty acid;

the gas chromatographic analysis conditions are that the temperatures of a Trace-TR-FAME column capillary column, a sample inlet and a detector are both 220 ℃; the sample injection amount is 1 mu L, the carrier gas is high-purity nitrogen, a constant flow mode with the split flow ratio of 1:100 is adopted, and the carrier gas flow rate is 1 mL/min.

7. The method of claim 1, wherein said collecting said liquid chromatogram of said experimental sample as tocopherol-based data in (S2) comprises: extracting tocopherol from the experimental sample, carrying out liquid chromatography analysis on the tocopherol to obtain a corresponding tocopherol map, and acquiring tocopherol basic data;

the extracting the tocopherol of the experimental sample comprises: adding the experimental sample adulterated with the tea oil into normal hexane, and dissolving and extracting to obtain tocopherol;

the conditions of the liquid chromatography are as follows: sehperisorb Silica column, column temperature 25 ℃, detection wavelength 295nm, sample injection amount 20 mu L, mobile phase of n-hexane, isopropanol (v: v ═ 98.5:1.5), flow rate 1 mL/min.

8. The method of claim 4, wherein the PLS-DA and/or SIMCA discriminatory models are optimized step by step (S5) by first modeling a adulteration ratio of 5% -100% and then successively increasing to 10% -100%, 20% -100%, 30% -100%, 40% -100%.

9. An apparatus or system, comprising:

a receiving unit configured to receive at least a portion of the fatty acid base data, the tocopherol base data of the experimental sample spiked with tea oil;

a memory for storing a PLS-DA discriminatory model and/or a SIMCA discriminatory model capable of generating a prediction of adulterated oil in the experimental sample of adulterated tea oil; and

a processor configured to access the PLS-DA and/or SIMCA discriminatory models stored in the memory to perform the steps of the method of any of claims 1-8 to generate a prediction of adulterated oil in the experimental sample of adulterated tea oil.

10. A computer readable storage medium comprising computer readable instructions which, when executed by a computer, operate to predict the authenticity of an adulterated oil adulterated with tea oil, the computer readable instructions being configured to perform the method of any one of claims 1 to 8.

Background

China is the country with the largest tea oil yield and has the name of 'east olive oil'. As a health-care nutritional oil, the tea oil has the same nutritional value as olive oil. Researches show that the tea oil contains not only high-content oleic acid, a small amount of linoleic acid, linolenic acid and the like, but also various physiological active substances such as squalene, sterol, tocopherol, tea polyphenol and other trace non-saponifiable substances. Oleic acid is used as monounsaturated fatty acid with the highest content in the tea oil, can obviously reduce harmful low-density lipoprotein cholesterol in blood and retain beneficial high-density lipoprotein cholesterol, and finally achieves the effect of reducing serum cholesterol. Due to the high nutritive value and economic value, other cheap grease can be added into the camellia oleosa seed oil by bad vendors in order to obtain the violence, and the camellia oleosa seed oil is sold as pure camellia oil, so that the quality of the camellia oleosa in the current market is uneven, and potential threats are caused to the health of consumers. Therefore, the quality and safety of the tea oil are guaranteed, which is a long-term and important subject.

In recent years, many researches successfully establish an advanced method for detecting the adulterated tea oil, and adopt advanced detection means such as an electronic nose, a differential scanning calorimetry method, an ion migration spectrum, a near/mid infrared spectrum, a Raman spectrum, a nuclear magnetic resonance spectrum, a fluorescence spectrum and the like. Although these modern techniques offer more options, chromatography, as a complementary targeting technique, remains a reliable method for identifying adulterated tea oil. By combining with chemometrics, chromatography is widely used in identifying vegetable oils, such as tea oil, olive oil, peanut oil, sesame oil, etc.

Chemometrics is the discipline that relates the measurements of a chemical system to the state of the system and is essentially the underlying theory and methodology of chemical metrology. The spectrogram data can be comprehensively analyzed by a chemometrics method, important information such as related research object components, structures and the like can be extracted from complex multidimensional data to the maximum extent, effective characteristic data can be obtained, a mathematical model can be established, interpretation, judgment and prediction of measurement data can be facilitated, and the method has the advantages of high calculation speed, good identification function and the like.

At present, no public report exists on a method for distinguishing adulterated tea oil doped with corn oil, rapeseed oil, rice bran oil, sesame oil and soybean oil by combining a fatty acid ratio obtained by gas chromatography and a tocopherol composition obtained by high performance liquid chromatography with chemometric methods such as HCA, PLS-DA, SIMCA and the like, and the technical field of the blank needs to be filled and perfected.

Disclosure of Invention

The purpose of the invention is as follows: in order to overcome the defects in the prior art, the invention provides a method for predicting the adulterated oil type in the tea oil, which comprises the steps of obtaining fatty acid maps of different oil samples through Gas Chromatography (GC), measuring tocopherol in the samples by adopting normal-phase High Performance Liquid Chromatography (HPLC), establishing the adulterated oil type in the tea oil by combining with chemometrics methods HCA, PLS-DA and SIMCA, and verifying the reliability of a model according to verification parameters and a verification set.

The technical scheme is as follows: in order to achieve the purpose, the invention adopts the technical scheme that:

a method for predicting the adulterated oil type in tea oil adopts a gas chromatography technology and a high performance liquid chromatography technology, combines chemometrics analysis to establish an identification model of the adulterated tea oil, and specifically comprises the following steps:

(S1) sample preparation: collecting qualified oil sold in different brands, taking tea oil as blended oil and other edible oil as adulterated oil, mixing the blended oil and the adulterated oil according to different volume proportions to obtain an experimental sample of the adulterated tea oil, wherein the experimental sample is divided into a training set and a verification set;

(S2) data acquisition: collecting the gas chromatogram of the experimental sample as fatty acid basic data, and collecting the liquid chromatogram of the experimental sample as tocopherol basic data;

(S3) unsupervised model building: based on unsupervised HCA analysis, respectively combining fatty acid basic data, tocopherol basic data and basic data of fatty acid and tocopherol to establish a corresponding pure oil distinguishing visual model;

(S4) supervised model building: taking fatty acid basic data and tocopherol basic data in a training set as input variables, and respectively combining with a supervised PLS-DA model and/or a supervised SIMCA model to establish a corresponding PLS-DA discrimination model and/or a supervised SIMCA discrimination model;

(S5) model verification: and respectively verifying the PLS-DA discrimination model and/or the SIMCA discrimination model by utilizing the fatty acid basic data and the tocopherol basic data which are centralized in verification to verify the accuracy of the models.

(S6) detecting application: processing a to-be-detected sample doped with the pseudo-tea oil, collecting gas chromatography of the to-be-detected sample as fatty acid detection data, collecting tocopherol detection data of liquid chromatography of the to-be-detected sample, and detecting the doping proportion of the to-be-detected sample doped with the pseudo-tea oil through a constructed PLS-DA discrimination model and/or a constructed SIMCA discrimination model.

In one embodiment, the fatty acid base data includes fatty acid ratio and the tocopherol base data includes tocopherol composition.

In one embodiment, the fatty acid ratios include P/S, O/P, O/L, O/Ln, and L/S; the tocopherol composition includes mass percentages of alpha-tocopherol, beta-tocopherol, gamma-tocopherol, and delta-tocopherol, and a ratio of gamma/alpha of gamma-tocopherol to alpha-tocopherol. P represents palmitic acid, S represents stearic acid, O represents oleic acid, L represents linoleic acid, Ln represents alpha-linolenic acid

In one embodiment, the adulteration ratio of the test sample of adulterated tea oil in (S1) is 0%, 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 100% (w/w), and the adulteration oil comprises any one of corn oil, rapeseed oil, rice bran oil, sesame oil, and soybean oil.

In one embodiment, the experimental sample of adulterated tea oil in (S1) is: optionally selecting 3 kinds of tea oil from 36 kinds of tea oil as sample tea oil, 3 kinds of corn oil, 3 kinds of rapeseed oil, 3 kinds of rice bran oil, 3 kinds of sesame oil and 3 kinds of soybean oil, and mixing with the three kinds of tea oil according to the proportion to obtain 135 adulterated samples, and 191 samples in total.

In one embodiment, the collecting the gas chromatogram of the experimental sample in (S2) as fatty acid basic data includes: extracting fatty acid of the experimental sample, carrying out gas chromatography on the fatty acid to obtain a corresponding fatty acid map, and acquiring basic data of the fatty acid;

the fatty acids from which the experimental sample is extracted include: carrying out methyl esterification on the experimental sample doped with the pseudo tea oil, and carrying out microfiltration on supernate to obtain fatty acid;

the gas chromatographic analysis conditions are that the temperatures of a Trace-TR-FAME column capillary column, a sample inlet and a detector are both 220 ℃; the sample injection amount is 1 mu L, the carrier gas is high-purity nitrogen, a constant flow mode with the split flow ratio of 1:100 is adopted, and the carrier gas flow rate is 1 mL/min.

In one embodiment, the collecting the liquid chromatogram of the experimental sample as the tocopherol basic data in (S2) comprises: extracting tocopherol from the experimental sample, carrying out liquid chromatography analysis on the tocopherol to obtain a corresponding tocopherol map, and acquiring tocopherol basic data;

the extracting the tocopherol of the experimental sample comprises: adding the experimental sample adulterated with the tea oil into normal hexane, and dissolving and extracting to obtain tocopherol;

the conditions of the liquid chromatography are as follows: sehperisorb Silica column, column temperature 25 ℃, detection wavelength 295nm, sample injection amount 20 mu L, mobile phase of n-hexane, isopropanol (v: v ═ 98.5:1.5), flow rate 1 mL/min.

In one embodiment, (S5) the PLS-DA model and/or the SIMCA model are gradually optimized by modeling the adulteration ratio of 5% -100% and then gradually increasing the adulteration ratio to 10% -100%, 20% -100%, 30% -100% and 40% -100%.

In another aspect, the present invention also provides an apparatus or system comprising:

a receiving unit configured to receive at least a portion of the fatty acid base data, the tocopherol base data of the experimental sample spiked with tea oil;

a memory for storing a PLS-DA model and/or a SIMCA model capable of generating a prediction of adulterated oil in the experimental sample of adulterated tea oil; and

a processor configured to access the PLS-DA model and/or SIMCA model stored in the memory to perform the steps of the method described above to generate a prediction of adulterated oil in the experimental sample of adulterated tea oil.

In another aspect, the present invention also provides a computer readable storage medium comprising computer readable instructions which, when executed by a computer, operate to predict the authenticity of an adulterated oil adulterated with tea oil, the computer readable instructions being configured to perform the method of any of the above.

Has the advantages that: compared with the prior art, the method for predicting the adulterated oil type in the tea oil has the following advantages:

(1) vegetable oils such as tea oil (CAO), corn oil (COO), rapeseed oil (RAO), Rice Bran Oil (RBO), sesame oil (SEO), and soybean oil (SOO) can be distinguished by fatty acid ratio, tocopherol composition, in combination with HCA, where all tea oil samples are better clustered together due to the higher oleic acid/alpha-linolenic acid ratio, oleic acid/palmitic acid ratio, oleic acid/linoleic acid ratio, and alpha-tocopherol content in the tea oil.

(2) The method is simple to operate, has reliable technology, and can quickly detect the adulterated oil type in the tea oil. The adulterated tea oil can be well identified through a PLS-DA discrimination model and a SIMCA discrimination model, the total discrimination accuracy rate reaches over 85.71 percent, the majority of discrimination accuracy rates are higher than 90 percent, and the highest discrimination accuracy rate reaches 97.73 percent.

(3) By comparing the discrimination accuracy of the PLS-DA discrimination model and the SIMCA discrimination model, the PLS-DA discrimination model is more suitable for discriminating high-concentration adulterated tea oil (the adulteration ratio is more than or equal to 20 percent when the PLS-DA discrimination model is 100 percent), and the SIMCA discrimination model is more suitable for discriminating low-concentration adulterated tea oil (the adulteration ratio is more than or equal to 5 percent).

Drawings

FIG. 1 is a heat map of fatty acid and tocopherol ratios, wherein (a) is a heat map of fatty acid ratios, (b) is a heat map of tocopherol ratios, and (c) is a heat map of fatty acid and tocopherol ratios. 1-toc% of alpha-tocopherol; 2-toc% of beta-tocopherol; 3-toc% of gamma-tocopherol; 4-toc%, delta-tocopherol%; 3-toc/1-toc is gamma-/alpha-tocopherol.

FIG. 2 is a PLS-DA score chart of tea oil samples of different adulteration types and adulteration proportions: (a) not less than 5 percent, (b) not less than 10 percent, (c) not less than 20 percent, (d) not less than 30 percent and (e) not less than 40 percent; CAO: tea oil, COO: corn oil, RAO: rapeseed oil, RBO: rice bran oil, SEO: sesame oil, SOO: and (3) soybean oil.

FIG. 3 is a SIMCA score chart of tea oil samples with different adulteration types and adulteration proportions: (a) not less than 5 percent, (b) not less than 10 percent, (c) not less than 20 percent, (d) not less than 30 percent and (e) not less than 40 percent; CAO: tea oil, COO: corn oil.

Detailed Description

The invention is further described with reference to the following figures and examples. The present invention will be better understood from the following examples. However, those skilled in the art will readily appreciate that the specific material ratios, process conditions and results thereof described in the examples are illustrative only and should not be taken as limiting the invention as detailed in the claims.

A specific embodiment of the present invention comprises the steps of:

(S1) sample preparation: collecting qualified oil sold in different brands, taking tea oil as blended oil and other edible oil as adulterated oil, and mixing the blended oil and the adulterated oil according to different volume proportions to obtain an experimental sample of the adulterated tea oil; before a supervision mode identification model, namely a PLS-DA discrimination model and/or a SIMCA discrimination model, is established, collected experimental samples are divided into a training set and a verification set;

in the present invention, the "PLS-DA discriminant model and/or the SIMCA discriminant model" includes three cases: first, a PLS-DA discriminative model; second, SIMCA discriminative model; third, PLS-DA discriminant model and SIMCA discriminant model. Three cases are within the scope of the present invention, namely, the PLS-DA discriminant model and the SIMCA discriminant model can be used individually or both. In this embodiment, two discriminant models are given corresponding examples for illustration, and each separately used PLS-DA discriminant model example is a complete example and can be used as a complete discriminant result; the results of the two discrimination models may be compared in the transverse direction, and the discrimination result of one discrimination model with a higher discrimination accuracy may be selected.

In one example, 2/3 random experimental samples are divided into training sets for establishing PLS-DA discriminant models and/or SIMCA discriminant models; the remaining 1/3 experimental samples serve as validation sets for evaluating the predictive power of the PLS-DA and/or SIMCA discriminatory models;

in one example, the types and amounts of oils used for the oil blended and the adulterated oil are respectively as follows: 36 kinds of tea oil, 6 kinds of corn oil, 3 kinds of rapeseed oil, 4 kinds of rice bran oil, 3 kinds of sesame oil and 4 kinds of soybean oil, and 56 kinds of pure vegetable oil samples are obtained in total; the total volume of mixed oil samples used for adulteration of the tea oil is 1mL, the adulteration proportion is 0%, 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 100% (w/w), 3 tea oil selected from 36 tea oil are used as sample tea oil, 3 corn oil, 3 rapeseed oil, 3 rice bran oil, 3 sesame oil and 3 soybean oil selected from the adulteration oil are selected and mixed with the three tea oil according to the proportion to obtain 135 adulteration samples, and the total volume of 191 samples.

The 191 samples are divided into 128 training sets and 63 verification sets, wherein the pure tea oil samples in the training sets and the samples doped with corn oil, rapeseed oil, rice bran oil, sesame oil and soybean oil are 24, 22, 20, 21, 20 and 21 in sequence, and the verification sets are 12, 11, 10 and 10 in sequence.

In one example, the PLS-DA and SIMCA discriminative models are gradually optimized by increasing the adulteration ratio by 5% -100% to 10% -100%, 20% -100%, 30% -100%, 40% -100%.

(S2) data acquisition: collecting the gas chromatogram of the experimental sample as fatty acid basic data, and collecting the liquid chromatogram of the experimental sample as tocopherol basic data;

in one example, the fatty acid base data includes fatty acid ratio and the tocopherol base data includes tocopherol composition. In one specific example, the fatty acid ratio values include P/S, O/P, O/L, O/Ln, and L/S; the tocopherol composition includes mass percentages of alpha-tocopherol, beta-tocopherol, gamma-tocopherol, and delta-tocopherol, and a ratio of gamma/alpha of gamma-tocopherol to alpha-tocopherol.

In one example, the fatty acid basic data and the tocopherol basic data are respectively acquired as follows:

s21 determination of fatty acid ratio

A: sample pretreatment: weighing about 50mg of sample into a 5mL centrifuge tube, adding 2mL of n-hexane and 0.5mL of 2mol/L KOH-CH₃OH solution; vortex for 30s, stand, take 1mL supernatant through 0.22 μm microporous membrane filtration and put in gas bottle for gas chromatography analysis.

B: gas chromatography conditions were a Trace-TR-FAME column capillary column (60 m.times.2.5 mm. times.2.5 mm, 0.20 μm, Varian Inc., USA); the temperature of the sample inlet and the detector are both 220 ℃; the sample injection amount is 1 mu L, the carrier gas is high-purity nitrogen, a constant flow mode with the split flow ratio of 1:100 is adopted, and the carrier gas flow rate is 1 mL/min. The temperature-raising program is as follows: from an initial temperature of 60 deg.C (for 3min) at a rate of 5 deg.C/min to 175 deg.C (for 15min), then at a rate of 2 deg.C/min to 220 deg.C (for 10min), and finally at a rate of 2 deg.C/min to 215 deg.C (for 20 min).

S22 determination of tocopherol composition

A: sample pretreatment: accurately weighing 100mg of oil sample, adding 1mL of n-hexane for dissolving, uniformly mixing by vortex, passing through a 0.22 mu m organic membrane, and carrying out liquid chromatography analysis.

B: chromatographic conditions are as follows: the tocopherol content in the oil sample is determined by normal phase high performance liquid chromatography, a Sehperisorb Silica column (25cm multiplied by 4.6mm multiplied by 5 mu m) is adopted, the column temperature is 25 ℃, the detection wavelength is 295nm, 20 mu L of sample injection is adopted, the mobile phase is normal hexane: isopropanol (v: v ═ 98.5:1.5), and the flow rate is 1 mL/min.

C: qualitative and quantitative analysis: and (3) measuring the contents of 4 configuration tocopherols including alpha, beta, gamma and delta-tocopherol and the total tocopherol content in the camellia seed oil by an external standard method.

(S3) unsupervised model building: based on unsupervised HCA analysis, respectively combining fatty acid basic data, tocopherol basic data and basic data of fatty acid and tocopherol to establish a corresponding pure oil distinguishing visual model;

(S4) supervised model building: taking fatty acid basic data and tocopherol basic data in a training set as input variables, and respectively combining a supervised PLS-DA model and an supervised SIMCA model to establish a PLS-DA discrimination model and an SIMCA discrimination model;

in a specific example, the obtained fatty acid ratio and tocopherol ratio data are introduced into MetaboAnalyst 5.0 and SIMCA 13(Umetrics, Malmo, Sweden) software to establish a PLS-DA discriminant model of cluster analysis (HCA) and partial least squares discriminant analysis and a SIMCA discriminant model of cluster independent soft mode method.

(S5) model verification: and respectively verifying the PLS-DA discrimination model and the SIMCA discrimination model by using the fatty acid basic data and the tocopherol basic data which are concentrated in verification, and verifying the accuracy of the models.

In one example, the method further comprises the steps of adjusting and setting model parameters, formulating a judgment rule, setting the tea oil value as 1, the adulterated tea oil value as 0, setting the threshold value as 0.5 in discriminant analysis, and specifying that more than or equal to 0.5 in predicted values is judged as 1, less than 0.5 is judged as 0, and obtaining the discrimination accuracy data of the PLS-DA discriminant model and the SIMCA discriminant model.

(S6) detecting application: processing a to-be-detected sample doped with the pseudo-tea oil, collecting gas chromatography of the to-be-detected sample as fatty acid detection data, collecting tocopherol detection data of liquid chromatography of the to-be-detected sample, and detecting the doping proportion of the to-be-detected sample doped with the pseudo-tea oil through a constructed PLS-DA discrimination model and a constructed SIMCA discrimination model.

In the above steps, this embodiment adopts the stoichiometric methods of HCA, PLS-DA, and SIMCA, and the specific relevant principles are as follows:

the term "clustering analysis (HCA)" refers to a method of determining the relationship between samples according to the similarity or difference of the samples by mathematical methods according to the attributes of the samples to be measured, and then clustering the samples to be measured according to the relationship. Generally speaking, the same type is used, that is, the samples to be measured in the same type are closer in the established model, while the samples to be measured in different types are farther away.

Partial least squares discriminant analysis (PLS-DA) is a supervised pattern recognition method, has strong capability of providing information, and the established model is more stable and has strong anti-interference capability, thus being a quantitative modeling method with the most extensive application. When the noise ratio of the tested sample is larger and multiple collinearity exists, the partial least square method is suitable for analyzing and identifying the experimental sample, and the obtained result can be more accurate.

The SIMCA (Soft independent modeling of class algorithms) is a supervised pattern recognition and classification method based on principal-component analysis (PCA) proposed by Wold in 1976, also called as a similarity analysis method. The basic idea of the algorithm is to establish a PCA model for each category by using the prior classification knowledge, and then judge the attribution of unknown samples by using the established models. The SIMCA method sets a confidence interval of classification by F-test, and for each class, the degrees of freedom of two dimensions of the F-test are respectively: (M-A) and (n-A-1) (M-A), wherein M is the number of variables (the number of variables should be the same for each class), A is the number of valid principal components of the class, and n is the number of samples of the class.

However, at present, no experimental basis is provided for the application of chromatographic technology combined with a chemometric method in the identification of adulteration of tea oil, wherein the fatty acid fingerprint obtained by gas chromatography and the tocopherol fingerprint obtained by high performance liquid chromatography are combined with HCA, PLS-DA and SIMCA to predict the types of other doped oils in the tea oil, including corn oil, rapeseed oil, rice bran oil, sesame oil and soybean oil.

Therefore, the invention provides a method for predicting the adulterated oil type in the tea oil, which comprises the steps of obtaining fatty acid maps of different oil samples through Gas Chromatography (GC), measuring tocopherol in the samples by adopting normal-phase High Performance Liquid Chromatography (HPLC), establishing the adulterated oil type in the tea oil by combining with chemometrics HCA, PLS-DA and SIMCA, and verifying the reliability of a model according to verification parameters and a verification set.

The following examples are given to further illustrate embodiments of the present invention:

in the following specific examples, the terms "concentration", "adulteration amount", "adulteration ratio" and the like used to describe the concentrations and ratios are all intended to mean: the mass percentage concentration of adulterated oil in the adulterated oil.

Example 1 sample configuration and data acquisition

1.1 this example tests fatty acids in tea oil, corn oil, rapeseed oil, rice bran oil, sesame oil and soybean oil according to flash methyl esterification and gas chromatography to obtain compositions having fatty acid ratios including P/S, O/P, O/L, O/Ln, and L/S, as shown in Table 1.

1.2 this example tests the tocopherol in tea oil, corn oil, canola oil, rice bran oil, sesame oil, and soybean oil according to normal phase high performance liquid chromatography to obtain the percent compositions of alpha, beta, gamma, and delta-tocopherols, and the gamma-tocopherol/alpha-tocopherol ratio, as shown in table 1.

TABLE 1 fatty acid and tocopherol composition in different vegetable oils

Example 2 unsupervised model building

Based on the fatty acid ratio, tocopherol composition, HCA analysis was performed on 56 pure vegetable oils: the 56 pure oil samples were best distinguished in fig. 1(c), where the tea oil samples were better clustered due to the higher oleic acid/alpha-linolenic acid ratio, oleic acid/palmitic acid ratio, oleic acid/linoleic acid ratio, and alpha-tocopherol content of the tea oil.

Example 3 supervised model establishment: PLS-DA discriminant model

And modeling and analyzing the adulterated oil types by combining a PLS-DA discrimination model according to the ratio of fatty acid and the composition of tocopherol.

As can be seen from FIG. 2 and Table 2, when the adulteration ratio is greater than or equal to 5%, the total discrimination accuracy is greater than or equal to 85.71%; with the increase of the adulteration ratio, the total recognition rate is gradually increased.

When the adulteration ratio is more than or equal to 20 percent, the total discrimination accuracy is more than or equal to 94.34 percent through a PLS-DA discrimination model, and the tea oil (CAO), the tea oil (CAO + COO) doped with a corn oil sample, the tea oil (CAO + SOO) doped with a soybean oil sample, the tea oil (CAO + RBO) doped with a rice bran oil sample, the tea oil (CAO + SEO) doped with a sesame oil sample and the tea oil (CAO + RAO) doped with a rapeseed oil sample can be better discriminated.

When the adulteration ratio is more than or equal to 40%, the total discrimination accuracy is more than or equal to 97.67% through a PLS-DA discrimination model, and the tea oil (CAO), the tea oil (CAO + COO) doped with a corn oil sample, the tea oil (CAO + SOO) doped with a soybean oil sample, the tea oil (CAO + RBO) doped with a rice bran oil sample, the tea oil (CAO + SEO) doped with a sesame oil sample and the tea oil (CAO + RAO) doped with a rapeseed oil sample can be distinguished more accurately.

From the above, it is demonstrated that the PLS-DA discrimination model can be used for better identifying the adulterated tea oil, and is particularly suitable for identifying the adulterated tea oil with high adulterated concentration (the adulteration ratio is more than or equal to 20%).

TABLE 2 discrimination accuracy of PLS-DA discrimination model

Note: more than or equal to 5% means: the adulteration amount is more than or equal to 5 percent when the adulteration amount is more than 100 percent;

more than or equal to 10% means: the adulteration amount is more than or equal to 10 percent when the adulteration amount is more than 100 percent;

not less than 20% means: the adulteration amount is more than or equal to 20 percent when the adulteration amount is more than 100 percent;

more than or equal to 30% means: the adulteration amount is more than or equal to 30 percent when the adulteration amount is more than 100 percent;

not less than 40% means: the adulteration amount is more than or equal to 40 percent when the adulteration amount is more than 100 percent;

the same applies to tables 3 and 4 below.

Example 4 supervised model establishment: SIMCA discrimination model

And (4) modeling and analyzing the adulterated oil types by combining a SIMCA (simple in analogy) discrimination model according to the fatty acid ratio and the tocopherol composition.

As can be seen from FIG. 3 and Table 3, when the adulteration ratio is greater than or equal to 5%, the CAO + COO, CAO + RAO, CAO + RBO, CAO + SEO and CAO + SOO can be better distinguished through the SIMCA distinguishing model, the distinguishing accuracy of the training set is more than 90%, the distinguishing accuracy of the training set of the pure tea oil is 91.67%, and the total distinguishing accuracy is greater than or equal to-92.06%.

When the adulteration ratio is more than or equal to 20 percent, the total discrimination accuracy is more than 92.45 percent through a SIMCA discrimination model, and the tea oil (CAO), the tea oil (CAO + COO) doped with the corn oil sample, the tea oil (CAO + SOO) doped with the soybean oil sample, the tea oil (CAO + RBO) doped with the rice bran oil sample, the tea oil (CAO + SEO) doped with the sesame oil sample and the tea oil (CAO + RAO) doped with the rapeseed oil sample can be better distinguished. And when the adulteration ratio is more than or equal to 30 percent, the number of samples of the verification set is further reduced, the accuracy of the verification set of the tea oil (CAO + COO) doped with the corn oil sample and the tea oil (CAO + RAO) doped with the rapeseed oil sample is lower, and the recognition rate of the total verification set is the lowest (89.58 percent).

From the above, it is demonstrated that the adulterated tea oil can be identified better by the SIMCA discrimination model, and the method is suitable for identifying the adulterated tea oil with different adulteration concentrations, wherein the adulteration proportion is more than or equal to 5%.

TABLE 3 discrimination accuracy of SIMCA discrimination model

The accuracy of the two discrimination models obtained in examples 3 and 4 was compared, and the results are shown in table 4.

Firstly, when the adulteration ratio is more than or equal to 20%, the discrimination accuracy of the PLS-DA discrimination model is more than 94.34%, and under the same adulteration ratio, the discrimination accuracy of the PLS-DA discrimination model is obviously higher than that of the SIMCA discrimination model (92.45%). Meanwhile, when the adulteration ratio is more than or equal to 30% and the adulteration ratio is more than or equal to 40%, the same result as that when the adulteration ratio is more than or equal to 20% can be found, namely, the recognition rate of the PLS-DA discrimination model is higher than that of the SIMCA discrimination model.

Therefore, it can be seen that when the adulteration ratio is greater than or equal to 20%, the PLS-DA discrimination model has higher discrimination accuracy (up to 97.67%) compared with the SIMCA discrimination model, so that the PLS-DA discrimination model is more suitable for identifying high-concentration adulterated tea oil (the adulteration ratio is greater than or equal to 20%).

And secondly, when the adulteration ratio is more than or equal to 10%, the discrimination accuracy of the SIMCA discrimination model is above 93.10%, and under the same adulteration ratio, the discrimination accuracy of the SIMCA discrimination model is obviously higher than that of the PLS-DA discrimination model (91.38%). Meanwhile, when the adulteration ratio is more than or equal to 5%, the same result as that when the adulteration ratio is more than or equal to 10% can be found, namely, the recognition rate of the SIMCA discrimination model is higher than that of the PLS-DA discrimination model.

Therefore, it can be seen that when the adulteration ratio is greater than or equal to 5%, compared with the PLS-DA discrimination model, the SIMCA discrimination model has higher discrimination accuracy (up to 96.88%), so that the SIMCA discrimination model is more suitable for identifying low-concentration adulterated tea oil (the adulteration ratio is greater than or equal to 5%).

By combining the above analysis, it can be seen that the PLS-DA discrimination model is more suitable for identifying high-concentration adulterated tea oil (adulteration ratio is more than or equal to 20%), and the SIMCA discrimination model is more suitable for identifying low-concentration adulterated tea oil (adulteration ratio is more than or equal to 5%). When the adulteration ratio of the adulterated tea oil is higher (the adulteration ratio is more than or equal to 20%), the judgment result of the PLS-DA discrimination model is preferably taken as the main reference, and when the adulteration ratio of the adulterated tea oil is lower (the adulteration ratio is more than or equal to 5%), the judgment result of the SIMCA discrimination model is preferably taken as the main reference.

TABLE 4 comparison of discrimination accuracy between PLS-DA discrimination model and SIMCA discrimination model

In conclusion, the invention provides a method for predicting the adulterated oil type in the tea oil, which comprises the steps of adding corn oil, rapeseed oil, rice bran oil, sesame oil and soybean oil in different proportions into the tea oil, and carrying out Gas Chromatography (GC) and High Performance Liquid Chromatography (HPLC) detection to obtain corresponding fatty acid and tocopherol maps. Tea oil, corn oil, rapeseed oil, rice bran oil, sesame oil, and soybean oil can be better distinguished by fatty acid ratio, tocopherol composition, in combination with HCA analysis. And then, the adulterated tea oil is judged through a PLS-DA (partial least squares-data acquisition) judging model and/or a SIMCA (simple least squares analysis and classification) judging model, the total judging accuracy rate reaches more than 85.71, the majority of judging accuracy rates are higher than 90%, and the highest judging accuracy rate reaches 97.73%. The method is simple to operate, has reliable technology, and can quickly detect the adulterated oil type in the tea oil.

When the adulteration amount is more than 20 percent (100 percent is larger than the adulteration ratio is more than or equal to 20 percent), the tea oil and the adulterated tea oil can be better distinguished through a PLS-DA distinguishing model, the distinguishing accuracy is more than 94.34 percent, and the highest distinguishing accuracy is up to 97.67 percent.

When the adulteration amount is more than or equal to 5 percent, the adulteration amount is combined with a SIMCA discrimination model, the adulterated tea oil can be better distinguished, the discrimination accuracy is more than 92.06 percent, and the highest discrimination accuracy is up to 96.88 percent.

The above description is only of the preferred embodiments of the present invention, and it should be noted that: it will be apparent to those skilled in the art that various modifications and adaptations can be made without departing from the principles of the invention and these are intended to be within the scope of the invention.