Baijiu flavor substance difference research method based on spectrum library retrieval and analysis

Abstract

The invention is applied to the technical field of white spirit, and particularly discloses a white spirit flavor substance difference research method based on spectrum library retrieval analysis, which comprises the following steps: S1, selecting a sample; s2, a sample pretreatment step; s3, an instrument analysis step; s4, a spectrum library retrieval step; s5, a result verification step; s6, a data analysis step; and S7, a result discussion step. According to the white spirit flavor substance difference research method based on spectrum library retrieval and analysis, a white spirit sample is analyzed by utilizing a gas chromatography-mass spectrometry analysis method, the analysis speed is high, meanwhile, complex components in white spirit can be efficiently separated and identified, the white spirit analysis efficiency is improved, and before analysis, the analysis result is more accurate and accurate. According to the method, pretreatment such as dilution and impurity removal is carried out, the influence of impurities in Baijiu on analysis results is avoided, the analysis accuracy is improved, three groups of samples are arranged for the same Baijiu sample in the analysis process of the Baijiu sample, and the result accuracy is improved.

Description

Technical Field

[0001] The invention relates to the technical field of liquor, and in particular to a method for studying the differences of liquor flavor substances based on spectral library retrieval analysis. Background Art

[0002] Baijiu is an alcoholic beverage made from grains through processes such as steaming, fermentation and distillation. Baijiu has a long history, dating back thousands of years. It occupies an important position in China's food culture and social occasions. After a long process of development, different regions have formed their own unique baijiu brewing techniques and flavors. The different flavor substances in baijiu and the content of the flavor substances affect the taste of baijiu. Therefore, studying the differences in flavor substances in baijiu will help develop and improve the taste of baijiu and increase the audience range of baijiu.

[0003] At present, the research process of flavor substances in liquor is traditionally carried out by extracting and analyzing liquor solution in the laboratory to detect various substances in the liquor. However, the traditional analysis method has a slow experimental speed, and different flavor substances require different experimental methods, which increases the detection time of flavor substances. Summary of the invention

[0004] The purpose of the present invention is to provide a method for studying the differences in liquor flavor substances based on spectral library retrieval analysis, so as to solve the problem of slow experimental speed and long detection time in the process of detecting liquor flavor substances by traditional chemical methods proposed in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solution: a method for studying the differences in flavor substances of liquor based on spectral library retrieval analysis, comprising the following steps:

[0006] S1. Sample selection step: several representative Maotai-flavor liquors from the Chishui River Basin were selected as research samples;

[0007] S2, sample pretreatment step, pretreatment of the collected samples, dilution of the liquor to within the detection range of the analytical instrument, and removal of impurities from the liquor samples by filtering and centrifugation;

[0008] S3, instrumental analysis step, using a gas chromatography-mass spectrometry analysis method to obtain a mass spectrum of the liquor sample after pretreatment;

[0009] S4, a spectral library search step, based on the analyzed mass spectrum of the liquor sample, an algorithm is used to search in a known spectral library to retrieve the result with the highest similarity to the sample;

[0010] S5, result verification step, verifying the results retrieved from the spectral library, comparing the sample with the standard and performing infrared spectrum auxiliary verification to confirm the sample composition;

[0011] S6, data analysis step, qualitatively and quantitatively analyzing the flavor components analyzed in the liquor sample, and processing and analyzing the data using statistics;

[0012] S7, result discussion step, based on the data analysis structure, combined with the production processes of different Maotai-flavor liquors in the Chishui River Basin, to explore the differences in liquor flavor substances;

[0013] Preferably, in step S1, when selecting samples, the research samples are selected based on process production, typical Maotai-flavor liquors are selected in the Chishui River Basin, and 3 samples are selected for each Maotai-flavor liquor as parallel samples, which are marked as X1 samples, X2 samples and X3 samples in sequence, wherein the X1 samples and X2 samples are used as samples for instrument analysis in step S3, and the X3 samples are used as comparison samples for result verification in step S5.

[0014] Using the above technical solution, three samples of each Maotai-flavor liquor were selected for parallel experiments, and parallel experiments were used to reduce accidental errors in the analysis process.

[0015] Preferably, in the step S2, when the sample is pretreated, the liquor sample is diluted to dilute the alcohol content of the liquor sample to 15%, and the diluted liquor sample is filtered and centrifuged to remove impurities in the liquor sample.

[0016] By adopting the above technical solution, pre-treatment of liquor samples can prevent impurities in the samples from affecting the analysis results.

[0017] Preferably, in the step S3, the pretreated liquor sample is injected into a gas chromatograph-mass spectrometer, the parameters of the gas chromatograph-mass spectrometer are adjusted, and the mass spectrum of the liquor sample is analyzed, wherein the injected liquor samples are X1 sample and X2 sample, and the obtained mass spectra are respectively marked as mass spectrum No. Y1 and mass spectrum No. Y2.

[0018] By adopting the above technical solution, the mass spectrum of each component substance in the liquor sample can be quickly analyzed using a gas chromatography-mass spectrometer.

[0019] Preferably, the spectral library retrieval in step S4 adopts a retrieval method that combines a decision tree classification model with a machine learning algorithm. The decision tree classification model performs feature classification on the known spectral library to improve the mass spectrum retrieval speed. The machine learning algorithm searches and matches the mass spectrum No. Y1 and the mass spectrum No. Y2 of the liquor sample respectively to obtain the mass spectrum of the compound with the highest similarity to the mass spectrum No. Y1 and the mass spectrum No. Y2, and uses the retrieved mass spectrum to train the machine learning algorithm.

[0020] By adopting the above technical solution, the decision tree classification model and the machine learning algorithm are combined to quickly search and analyze the sample mass spectrum and the spectral library to obtain the compounds with the highest similarity to the flavor substances in the liquor sample.

[0021] Preferably, the method for generating the decision tree classification model is as follows: the known spectral library is marked as a total spectrum set, the total spectrum set is classified according to the spectral peak position, and after classification, it is marked as a first-generation subset, each of the first-generation subsets is classified again according to the spectral peak intensity, and after classification, it is marked as a second-generation subset, each of the second-generation subsets is classified according to the spectral peak width, and after classification, it is marked as a retrieval subset, and the generated decision tree classification model is trained using the spectra of known compounds.

[0022] By adopting the above technical solution and using the decision tree classification model to pre-classify the spectral library, the retrieval speed of the mass spectra of liquor samples is increased, and the analysis efficiency is improved.

[0023] Preferably, in step S5, the compound with the highest similarity to the X1 sample and the X2 sample retrieved from the spectral library is used as a standard, the X3 sample is compared with the standard, and the X3 sample is subjected to infrared spectroscopy-assisted analysis to determine the composition of the liquor sample.

[0024] By adopting the above technical solution, the composition of the sample can be determined by comparing the X3 sample with the standard.

[0025] Preferably, in step S6, the components and contents of flavor substances in the liquor sample are determined according to the spectral library retrieval step and the result verification step, so as to obtain the proportion of each flavor substance in different types of the liquor samples, and based on the proportion of flavor substances, the representative flavor substances in the liquor samples are extracted, and based on the differences in the representative flavor substances in different liquor samples and combined with the production process of the liquor samples, the differences in the flavor substances of the liquor in the Chishui River Basin are analyzed.

[0026] By adopting the above technical solution, the representative flavor substances in the liquor sample can be obtained using the proportion of flavor substances, and compared with the remaining samples to obtain the differences in the flavor substances of the liquor.

[0027] Compared with the prior art, the present invention has the following beneficial effects: the method for studying the differences in flavor substances of liquor based on spectral library retrieval analysis:

[0028] 1. The present invention uses a gas chromatography-mass spectrometry analysis method to analyze liquor samples. While the analysis speed is fast, the complex components in the liquor can be efficiently separated and identified, thereby improving the efficiency of liquor analysis. Before the analysis, pretreatment such as dilution and impurity removal is performed to avoid the impurities in the liquor affecting the analysis results, thereby increasing the accuracy of the analysis.

[0029] Furthermore, in the process of analyzing liquor samples, three groups of samples of the same liquor samples were set, among which X1 sample and X2 sample were used for parallel experiments of mass spectrometry analysis, and the parallel experiment was used to reduce the accidental error of the experiment and improve the credibility of the analysis results. After the mass spectrum was retrieved, the X3 sample was used as an experimental sample for comparison with the standard to improve the accuracy of the results.

[0030] 2. In the present invention, a retrieval method combining a decision tree classification model and a machine learning algorithm is set in the spectral library retrieval step. The decision tree classification model is used to classify the known spectral library into different retrieval subsets according to the spectral peak position, spectral peak intensity and spectral peak width. The spectral features in different retrieval subsets are different. According to the generated decision tree classification model, the liquor sample spectrum can be quickly classified into a certain retrieval subset for retrieval, thereby improving the retrieval efficiency;

[0031] Furthermore, the classified liquor sample spectra are used to retrieve the retrieval subset using a machine learning algorithm. The retrieval results can be fed back into the machine learning algorithm using the machine learning algorithm to train the machine learning algorithm and improve the retrieval efficiency and accuracy of the machine learning algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a schematic diagram of the process structure of the present invention;

[0033] Figure 2 This is a schematic diagram of the structure of the decision tree classification model of the present invention;

[0034] Figure 3 It is a schematic diagram of the structure of the sample analysis results of the present invention. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0036] See also Figure 1-Figure 3 The present invention provides a technical solution: a method for studying the differences in flavor substances in liquor based on spectral library retrieval analysis.

[0037] The method comprises the following steps: S1, a sample selection step, in which several representative Maotai-flavor liquors in the Chishui River Basin are selected as research samples; S2, a sample pretreatment step, in which the collected samples are pretreated, the liquor is diluted to within the detection range of the analytical instrument, and the liquor samples are subjected to impurity removal treatment by filtering and centrifugation; S3, an instrument analysis step, in which the pretreated liquor samples are subjected to gas chromatography-mass spectrometry analysis to obtain a mass spectrum of the liquor samples; S4, a spectral library search step, in which the mass spectrum of the liquor samples analyzed is searched in a known spectral library The algorithm is used to search and retrieve the result with the highest similarity to the sample; S5, the result verification step, verifies the result retrieved from the spectral library, compares the sample with the standard and performs infrared spectrum auxiliary verification to confirm the sample composition; S6, the data analysis step, conducts qualitative and quantitative analysis on the flavor components analyzed in the liquor sample, and uses statistics to process and analyze the data; S7, the result discussion step, based on the data analysis structure, combined with the production processes of different sauce-flavor liquors in the Chishui River Basin, discusses the differences in liquor flavor substances;

[0038] like Figure 1 As shown, in the process of studying the differences in flavor substances of liquor, samples of liquor from the Chishui River Basin were first selected. During the selection process, a variety of representative sauce-flavor liquors with different production processes and brands in the Chishui River Basin were selected as research samples. After the sample selection was completed, the collected samples were pretreated, the liquor samples were diluted, and filtered and centrifuged to remove impurities to prevent impurities from affecting the analysis results. The treated liquor samples were injected into a gas chromatography-mass spectrometer to obtain a mass spectrum of the liquor sample, and the mass spectrum of the liquor sample was input into a known spectral library for retrieval to obtain the result with the highest similarity. The result was verified with the liquor sample to confirm the sample composition, and the flavor substances in the liquor sample were qualitatively and quantitatively analyzed to obtain representative flavor substances of different types of sauce-flavor liquors, and different samples were compared to obtain the differences in flavor substances of sauce-flavor liquors.

[0039] In step S1, when selecting samples, the research samples are selected based on process production. Typical sauce-flavor liquors are selected in the Chishui River Basin. Three samples of each sauce-flavor liquor are selected as parallel samples. The three parallel samples are marked as X1 samples, X2 samples and X3 samples in turn, among which X1 samples and X2 samples are used as samples for instrument analysis in step S3, and X3 samples are used as comparison samples for result verification in step S5. In step S2, when pre-treating the samples, the liquor samples are diluted to dilute the alcohol content of the liquor samples to 15%. The diluted liquor samples are filtered and centrifuged to remove the liquor samples. In step S3, the pretreated liquor sample is injected into the gas chromatography-mass spectrometer, the parameters of the gas chromatography-mass spectrometer are adjusted, and the mass spectra of the liquor samples are analyzed, wherein the injected liquor samples are X1 sample and X2 sample, and the obtained mass spectra are marked as mass spectra Y1 and Y2, respectively. The spectral library retrieval in step S4 adopts a retrieval method combining a decision tree classification model and a machine learning algorithm. The decision tree classification model performs feature classification on the known spectral library to improve the mass spectrum retrieval speed, and the machine learning algorithm performs feature classification on the mass spectra Y1 and Y2 of the liquor samples, respectively. A search and match is performed to obtain the mass spectrum of the compound with the highest similarity to the mass spectrum of Y1 and the mass spectrum of Y2, and the retrieved mass spectrum is used to train the machine learning algorithm. The generation method of the decision tree classification model is as follows: the known spectral library is marked as a total spectrum set, the total spectrum set is classified according to the peak position, and after classification, it is marked as a first-generation subset, each first-generation subset is classified again according to the peak intensity, and after classification, it is marked as a second-generation subset, each second-generation subset is classified according to the peak width, and after classification, it is marked as a search subset. The generated decision tree classification model is trained using the spectra of known compounds. In step S5, the spectral library is retrieved. The compound with the highest similarity to the X1 sample and the X2 sample is used as a standard, and the X3 sample is compared with the standard, and the X3 sample is subjected to infrared spectrum auxiliary analysis to determine the composition of the liquor sample. In step S6, the flavor component and content in the liquor sample are determined according to the library search step and the result verification step, so as to obtain the proportion of each flavor substance in different types of liquor samples. According to the proportion of flavor substances, the representative flavor substances in the liquor sample are extracted. According to the differences in the representative flavor substances in different liquor samples, combined with the production process of the liquor samples, the differences in the flavor substances of the liquor in the Chishui River Basin are analyzed;

[0040] like Figure 2 and Figure 3As shown, when selecting samples, in the Chishui River Basin, the production process of liquor was used as the selection basis, and different brands of sauce-flavor liquor under different production processes were selected as samples for research. Three samples of different liquors were set, of which two X1 samples and X2 samples were used as parallel samples for mass spectrometry analysis to reduce accidental errors in the analysis process, and the other X3 sample was used as a result verification comparison sample. The X1 sample and the X2 sample were pretreated, and the X1 sample and the X2 sample were diluted to dilute the alcohol content to 15%. The diluted X1 sample and the X2 sample were filtered and centrifuged to remove impurities in the samples to avoid the influence of impurities on the analysis results. The X1 sample and the X2 sample were analyzed by gas chromatography-mass spectrometry respectively to obtain mass spectra Y1 and Y2. The mass spectra Y1 and Y2 were parallel groups. The parallel mass spectra Y1 and Y2 were used for mutual reference to obtain accurate analysis results. , reduce accidental errors, mass spectra Y1 and Y2 are input into the spectral library, and a decision tree classification model is built into the spectral library. The decision tree classification model classifies the known mass spectra in the spectral library by features, and classifies them in sequence according to the peak position, peak intensity and peak width to form a specific retrieval subset. After the mass spectra Y1 and Y2 are input, the machine learning algorithm searches the mass spectra of the samples in the corresponding retrieval subset to improve the mass spectrum retrieval speed, and uses the compound with the highest similarity retrieved from the spectral library as the standard to compare with the X3 sample. Infrared spectroscopy is used to assist in analysis to determine the flavor component and content of the liquor sample, thereby conducting a qualitative and quantitative analysis of the liquor sample, and according to the content proportion of different flavor substances, the representative flavor substances in the liquor sample are obtained, and the representative flavor substances are compared with different liquor samples. Combined with the production process of the liquor samples, the flavor substances of the liquor in the Chishui River Basin are differentially studied.

[0041] Working principle: Divide the liquor samples into three parallel sample groups, conduct parallel experiments of mass spectrometry analysis on sample X1 and sample X2, and conduct comparison experiments on sample X3 and standard products, so as to reduce accidental errors in the experimental process and improve the accuracy of the experimental results. After pretreatment of the samples, the mass spectrum of the samples is obtained by gas chromatography-mass spectrometry analysis, and the mass spectrum is input into the spectral library. The retrieval method combining the decision tree classification model with the machine learning algorithm is used to quickly retrieve the flavor substances and the content of flavor substances in the liquor samples. According to the content proportion of different flavor substances, the representative flavor substances are compared with different liquor samples. Combined with the production process of the liquor samples, the differences of the flavor substances in the liquor are studied.

[0042] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for studying the differences in flavor substances in liquor based on spectral library retrieval analysis, characterized in that: The following steps are involved: S1. Sample selection step: several representative Maotai-flavor liquors from the Chishui River Basin were selected as research samples; S2, sample pretreatment step, pretreatment of the collected samples, dilution of the liquor to within the detection range of the analytical instrument, and removal of impurities from the liquor samples by filtering and centrifugation; S3, instrumental analysis step, using a gas chromatography-mass spectrometry analysis method to obtain a mass spectrum of the liquor sample after pretreatment; S4, a spectral library search step, based on the analyzed mass spectrum of the liquor sample, an algorithm is used to search in a known spectral library to retrieve the result with the highest similarity to the sample; S5, result verification step, verifying the results retrieved from the spectral library, comparing the sample with the standard and performing infrared spectrum auxiliary verification to confirm the sample composition; S6, data analysis step, qualitatively and quantitatively analyzing the flavor components analyzed in the liquor sample, and processing and analyzing the data using statistics; S7. Result discussion step: Based on the data analysis structure and combined with the production processes of different Maotai-flavor liquors in the Chishui River Basin, the differences in liquor flavor substances were explored.

2. The method for studying the differences of flavor substances in liquor based on spectral library retrieval analysis according to claim 1, characterized in that: In step S1, when selecting samples, the research samples are selected based on process production, and typical Maotai-flavor liquors are selected in the Chishui River Basin. Three samples of each Maotai-flavor liquor are selected as parallel samples, and the three parallel samples are marked as X1 samples, X2 samples, and X3 samples in sequence, wherein the X1 samples and X2 samples are used as samples for instrument analysis in step S3, and the X3 samples are comparison samples for result verification in step S5.

3. The method for studying the differences of flavor substances in liquor based on spectral library retrieval analysis according to claim 1, characterized in that: In the step S2, when the sample is pretreated, the liquor sample is diluted to dilute the alcohol content of the liquor sample to 15%, and the diluted liquor sample is filtered and centrifuged to remove impurities in the liquor sample.

4. The method for studying the differences of flavor substances in liquor based on spectral library retrieval analysis according to claim 1, characterized in that: In the step S3, the pretreated liquor sample is injected into a gas chromatograph-mass spectrometer, the parameters of the gas chromatograph-mass spectrometer are adjusted, and the mass spectra of the liquor samples are analyzed, wherein the injected liquor samples are X1 sample and X2 sample, and the obtained mass spectra are marked as mass spectra Y1 and Y2, respectively.

5. The method for studying the differences of flavor substances in liquor based on spectral library retrieval analysis according to claim 1 is characterized by: The spectral library search in step S4 adopts a retrieval method that combines a decision tree classification model with a machine learning algorithm. The decision tree classification model performs feature classification on the known spectral library to improve the mass spectrum retrieval speed. The machine learning algorithm searches and matches the mass spectrum No. Y1 and the mass spectrum No. Y2 of the liquor sample respectively to obtain the mass spectrum of the compound with the highest similarity to the mass spectrum No. Y1 and the mass spectrum No. Y2, and uses the retrieved mass spectrum to train the machine learning algorithm.

6. The method for studying the differences of flavor substances in liquor based on spectral library retrieval analysis according to claim 5 is characterized by: The method for generating the decision tree classification model is as follows: the known spectral library is marked as a total spectrum set, the total spectrum set is classified according to the spectral peak position, and after classification, it is marked as a first-generation subset, each of the first-generation subsets is classified again according to the spectral peak intensity, and after classification, it is marked as a second-generation subset, each of the second-generation subsets is classified according to the spectral peak width, and after classification, it is marked as a retrieval subset, and the generated decision tree classification model is trained using the spectra of known compounds.

7. The method for studying the differences of flavor substances in liquor based on spectral library retrieval analysis according to claim 1, characterized in that: In the step S5, the compound with the highest similarity to the X1 sample and the X2 sample retrieved from the spectral library is used as a standard, the X3 sample is compared with the standard, and the X3 sample is subjected to infrared spectrum auxiliary analysis to determine the composition of the liquor sample.

8. The method for studying the differences of flavor substances in liquor based on spectral library retrieval analysis according to claim 1 is characterized by: In the step S6, the components and contents of flavor substances in the liquor samples are determined according to the spectral library retrieval step and the result verification step, so as to obtain the proportion of each flavor substance in different types of the liquor samples. Based on the proportion of flavor substances, representative flavor substances in the liquor samples are extracted. Based on the differences in representative flavor substances in different liquor samples and combined with the production process of the liquor samples, the differences in flavor substances of the liquor in the Chishui River Basin are analyzed.

Images