White spirit producing area identification method and device, computer equipment and storage medium

Abstract

The invention relates to a white spirit production area identification method and device, computer equipment and a storage medium, and the method comprises the steps: obtaining a first feature data set of to-be-identified white spirit, the first feature data set comprising flavor compound data, physicochemical index data and mineral element data; and determining the producing area of the to-be-identified white spirit based on the first feature data set. According to the method, the production area of the Baijiu is determined through the uniqueness of the Baijiu in different production areas reflected by three dimensions of the flavor compound data, the physicochemical index data and the mineral element data of the Baijiu, and the identification accuracy of the production area of the Baijiu can be remarkably improved.

Description

Technical Field

[0001] The present application relates to the technical field of liquor identification, and in particular to a method, device, computer equipment and storage medium for identifying liquor producing areas. Background Art

[0002] With the improvement of people's living standards and the continuous changes in consumption concepts, the liquor producing areas have gradually become an important basis for consumers to choose.

[0003] Liquor is a traditional alcoholic beverage with obvious regional characteristics. Affected by factors such as environment, climate, soil, water source, technology, and microorganisms, the quality characteristics of liquor in different brewing areas vary significantly. However, the identification of liquor producing areas currently relies mainly on sensory evaluation by professionals, and the results are easily affected by the physical and mental conditions of the evaluators, and there are problems such as strong subjectivity, difficulty in unifying the evaluation criteria, and poor repeatability. Summary of the invention

[0004] In order to solve the above technical problems, the present application provides a method, device, computer equipment and storage medium for identifying a liquor producing area.

[0005] In a first aspect, the present application provides a method for identifying a liquor producing area, the method comprising:

[0006] Acquire a first characteristic data set of liquor to be identified, wherein the first characteristic data set includes flavor compound data, physical and chemical index data, and mineral element data;

[0007] The production area of ​​the liquor to be identified is determined based on the first feature data set.

[0008] In one embodiment, the flavor compound data includes the types and contents of flavor compounds, and the types are detected based on at least one detection method of gas chromatography-flame ionization detector GC-FID, liquid-liquid microextraction-gas chromatography-mass spectrometry LLME-GC-MS and high performance liquid chromatography HPLC.

[0009] In one embodiment, obtaining the flavor compound data of the liquor to be identified includes:

[0010] Based on the GC-FID and internal standard curve method, obtaining the first flavor compound data of the liquor to be identified;

[0011] Based on the LLME-GC-MS and internal standard curve method, obtaining the second flavor compound data of the liquor to be identified;

[0012] Based on the HPLC and external standard curve method, obtaining the third flavor compound data of the liquor to be identified;

[0013] The target flavor compound data of the liquor to be identified is obtained according to the first flavor compound data, the second flavor compound data and the third flavor compound data.

[0014] In one embodiment, the conditions of the GC-FID include:

[0015] The model of the chromatographic column is DB-WAXUI, the specification of the chromatographic column is 30m×0.25mm×0.25μm, the injection port temperature and the hydrogen flame ionization detector temperature are both 250°C, the injection volume is 1μl, the split ratio is 30:1, the carrier gas is nitrogen (purity ≥99.999%), the column flow rate is 1ml / min, and the heating program is to maintain the initial temperature of 30°C for 3min, increase the temperature to 90°C at a heating rate of 3°C / min, and then increase the temperature to 180°C at a heating rate of 5°C / min, and maintain it for 25min;

[0016] and / or,

[0017] The gas chromatography GC conditions of the LLME-GC-MS include:

[0018] The model of the chromatographic column is DB-FFAP, and the specification of the chromatographic column is 60m×0.25mm×0.25μm. The injection port temperature and the gas chromatography detector temperature are both 250°C. Helium is used as the carrier gas (purity ≥99.999%), the flow rate is 1.2ml / min, and the injection is non-split. The temperature program is to maintain the initial temperature of 40°C for 1min, increase the temperature to 100°C at a heating rate of 2.5°C / min, maintain it for 1min, increase the temperature to 160°C at a heating rate of 3°C / min, and then increase the temperature to 230°C at a heating rate of 5°C / min, and maintain it for 10min.

[0019] The mass spectrometry MS conditions of the LLME-GC-MS include:

[0020] An EI ion source and an electron energy of 70 eV were used, the ion source temperature was set to 230 °C, the quadrupole temperature was set to 150 °C, and the scanning range was m / z 30 to 550;

[0021] and / or,

[0022] The HPLC conditions include:

[0023] The model of the chromatographic column is SB-C18, the specifications of the chromatographic column are 4.6×150mm, 5-Micron, the mobile phase is 100% 0.02mol / L KH2PO4, pH=2.3, the detection wavelength is 208nm, the column oven is 30°C, the mobile phase flow rate is 0.8ml / min, and the injection volume is 1μl.

[0024] In one embodiment, determining the production area of ​​the to-be-identified liquor based on the first feature data set includes:

[0025] Inputting the first feature data set into a liquor production area prediction model to obtain a predicted production area output by the liquor production area prediction model;

[0026] Among them, the liquor production area prediction model is trained based on a training data set and a machine learning model, the training data set includes a second feature data set of liquor samples from multiple production areas, the second feature data set includes flavor compound data samples, physical and chemical index data samples, and mineral element data samples, and the label corresponding to the second feature data set includes the production area of ​​the liquor sample corresponding to the second feature data set.

[0027] In one embodiment, after obtaining the predicted production area output by the liquor production area prediction model, the method further includes:

[0028] The contribution value of the first feature data set to the prediction result of the liquor production area prediction model is calculated to explain the prediction process of the liquor production area prediction model.

[0029] In one embodiment, before model training, the training data set is subjected to equalization processing;

[0030] And / or, the liquor production area prediction model is obtained based on XGBoost model training.

[0031] In one embodiment, the physical and chemical index data includes at least one of total acid content, total ester content and solid content;

[0032] And / or, the mineral element data is obtained based on inductively coupled plasma mass spectrometer ICP-MS detection.

[0033] In a second aspect, the present application provides a device for identifying a liquor producing area, the device comprising:

[0034] A data acquisition module, used to acquire a first characteristic data set of the liquor to be identified, wherein the first characteristic data set includes flavor compound data, physical and chemical index data, and mineral element data;

[0035] A production area recognition model is used to determine the production area of ​​the liquor to be identified based on the first feature data set.

[0036] In a third aspect, the present application provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for identifying a liquor producing area described in the first aspect is implemented.

[0037] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method for identifying a liquor producing area described in the first aspect is implemented.

[0038] On the basis of conforming to the common sense in the art, the above-mentioned preferred conditions can be arbitrarily combined to obtain the preferred embodiments of the present application.

[0039] The above-mentioned method, device, computer equipment and storage medium for identifying liquor production areas can achieve the following beneficial effects: determining the liquor production area by reflecting the uniqueness of liquor from different production areas through the three dimensions of liquor flavor compound data, physical and chemical index data and mineral element data, which can significantly improve the accuracy of identifying the liquor production area. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 A schematic diagram of a process for identifying a liquor producing area in one embodiment;

[0041] Figure 2 A schematic diagram of the training process of a liquor production area prediction model in one embodiment;

[0042] Figure 3 This is a schematic diagram showing the comparison of the training data set before and after equalization processing in Example 1;

[0043] Figure 4 This is a schematic diagram of the contribution value of the second feature data set to the prediction result of the first liquor production area prediction model in Example 1;

[0044] Figure 5 This is a schematic diagram showing the comparison of the training data set before and after equalization processing in Example 2;

[0045] Figure 6 This is a schematic diagram of the contribution value of the second feature data set to the prediction result of the second liquor production area prediction model in Example 2;

[0046] Figure 7 A schematic diagram of a module of a device for identifying a liquor producing area in one embodiment;

[0047] Figure 8 is a first internal structure diagram of a computer device in one embodiment;

[0048] Fig. 9 A second internal structure diagram of a computer device in one embodiment. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0050] It should be noted that the diagrams provided in the present embodiment only illustrate the basic concept of the present application in a schematic manner. The diagrams only show the components related to the present application rather than the number, shape and size of the components in the actual implementation. The form, quantity and proportion of each component in the actual implementation can be changed at will, and the component layout form may also be more complicated. The structure, proportion, size, etc. illustrated in the drawings of the present specification are only used to match the content disclosed in the specification for people familiar with this technology to understand and read, and are not used to limit the limiting conditions that can be implemented in the present application, so they have no technical substantive significance. Any modification of the structure, change of the proportional relationship or adjustment of the size, without affecting the effect that can be produced by the present application and the purpose that can be achieved, should still fall within the scope of the technical content disclosed in the present application. At the same time, the terms such as "upper", "lower", "left", "right", "middle" and "one" quoted in the present specification are only for the convenience of narration, and are not used to limit the scope of the implementation of the present application. The change or adjustment of the relative relationship, without substantial change of the technical content, should also be regarded as the scope of the implementation of the present application.

[0051] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various places in the text does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0052] As shown in this document, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural, unless the context clearly indicates an exception. Generally speaking, the terms "include" and "comprise" only indicate that the steps and elements that have been clearly identified are included, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0053] The definitions of inclusion in this document, such as the terms "having", "may have", "including" or "may include" used herein, indicate the existence of the corresponding functions, operations, elements, etc. herein, and do not limit the existence of one or more other functions, operations, elements, etc. In addition, it should be understood that the terms "including" or "having" used herein indicate the existence of the features, numbers, steps, operations, elements, components, or a combination thereof described in the specification, and do not exclude the existence or addition of one or more other features, numbers, steps, operations, elements, components, or a combination thereof.

[0054] In the embodiments of the present application, prefixes such as "first" and "second" are used only to distinguish different description objects, and have no limiting effect on the position, order, priority, quantity or content of the described objects. The use of prefixes such as ordinal numbers to distinguish description objects in the embodiments of the present application does not constitute a limitation on the described objects. For the statement of the described objects, please refer to the description in the context of the claims or embodiments, and no unnecessary limitation should be constituted due to the use of such prefixes. In addition, in the description of the present embodiment, unless otherwise specified, the meaning of "multiple" is two or more.

[0055] The identification method of liquor producing area provided in this application is as follows: Figure 1 As shown, steps S101-S102 are included:

[0056] S101, obtaining a first characteristic data set of liquor to be identified, the first characteristic data set including flavor compound data, physical and chemical index data, and mineral element data.

[0057] Specifically, step S101 includes:

[0058] Obtain the flavor compound data of the liquor to be identified.

[0059] Obtain the physical and chemical index data of the liquor to be identified;

[0060] Obtain the mineral element data of the liquor to be identified.

[0061] The flavor compound data include the type and content of the flavor compound.

[0062] In one embodiment, the types of flavor compounds can be detected based on at least one detection method of gas chromatography-flame ionization detector GC-FID, liquid-liquid microextraction-gas chromatography-mass spectrometry LLME-GC-MS and high performance liquid chromatography HPLC.

[0063] That is, it can be detected by any one of the detection methods, by combining any two of the detection methods, or by combining the above three detection methods.

[0064] Different detection methods can detect different flavor compounds. In order to improve the accuracy and comprehensiveness of flavor compound data, two or three detection methods can be used in combination to comprehensively determine the flavor compound data.

[0065] Exemplarily, the type sets of flavor compounds in the to-be-identified liquor are first detected based on different detection methods, and then the content of each flavor compound in each type set is determined separately, that is, partial flavor compound data of the to-be-identified liquor are detected based on different detection methods, and finally, the partial flavor compound data detected based on different detection methods are merged to obtain the target flavor compound data of the to-be-identified liquor.

[0066] In one embodiment, the three detection methods of GC-FID, LLME-GC-MS, and HPLC can be used in combination to qualitatively and quantitatively identify the flavor compounds in the liquor.

[0067] That is, the flavor compound data of the liquor to be identified is obtained, including:

[0068] Based on GC-FID and internal standard curve method, the first flavor compound data of the liquor to be identified is obtained;

[0069] Based on LLME-GC-MS and internal standard curve method, the data of the second flavor compound of the liquor to be identified are obtained;

[0070] Based on HPLC and external standard curve method, the third flavor compound data of the liquor to be identified is obtained;

[0071] According to the first flavor compound data, the second flavor compound data and the third flavor compound data, target flavor compound data of the liquor to be identified is obtained.

[0072] Exemplarily, based on GC-FID and internal standard curve method, obtaining the first flavor compound data of the liquor to be identified includes steps S201-S205:

[0073] S201, preparing a first sample to be tested based on the liquor to be identified.

[0074] For example, take 10 ml of the liquor to be identified and put it in a reserve bottle, and add 200 μl of the mixed internal standard preparation solution (tert-amyl alcohol: 297.46 mg / L, n-amyl acetate: 297.59 mg / L, 2-ethylbutyric acid: 302.37 mg / L) to the reserve bottle, mix well and take 2 ml to dispense into the injection bottle to obtain the first sample to be tested.

[0075] S202, using GC-FID to detect the first sample to be tested to obtain a first detection spectrum.

[0076] For example, conditions for GC-FID include:

[0077] The model of the chromatographic column is DB-WAXUI, the specification of the chromatographic column is 30m×0.25mm×0.25μm, the injection port temperature and the hydrogen flame ionization detector temperature are both 250°C, the injection volume is 1μl, the split ratio is 30:1, the carrier gas is nitrogen (purity ≥99.999%), and the column flow rate is 1ml / min;

[0078] The heating program was to maintain the initial temperature at 30°C for 3 min, increase the temperature to 90°C at a heating rate of 3°C / min, and then increase the temperature to 180°C at a heating rate of 5°C / min and maintain for 25 min.

[0079] S203, determining a first category set of flavor compounds in the liquor to be identified according to the first detection spectrum;

[0080] For example, after the test is completed, the flavor compounds detected in the liquor to be identified are qualitatively identified using NIST spectral library search, standard comparison, and retention index (RI).

[0081] The NIST spectral library contains a large number of mass spectra and corresponding RIs of known compounds. By comparing the first detection spectrum with the mass spectrum data in the NIST spectral library, the type of each flavor compound in the first detection spectrum can be determined.

[0082] Comparison of RIs can provide additional evidence to confirm the type of flavor compound. Even if the mass spectra of two compounds are very similar, the corresponding RIs may be significantly different. Therefore, the RI of each flavor compound to be identified in the first detection spectrum can also be calculated based on the information in the first detection spectrum, and the RI of the flavor compound to be identified can be compared with the RI of known compounds in the NIST spectral library to more accurately identify the type of flavor compound.

[0083] Standard comparison is to compare the chromatogram of the unknown sample with the chromatogram of the known standard. Under the same chromatographic operating conditions, by comparing the peak information of the known pure sample and the unknown substance, the unknown substance can be determined.

[0084] Comparison with standards is a direct qualitative method, while using the NIST spectral library search is a database-based auxiliary qualitative analysis method. Combining the two can further improve the accuracy and comprehensiveness of the types of flavor compounds.

[0085] S204, using an internal standard curve method to quantify each flavor compound in the first category set to obtain a first content set of flavor compounds.

[0086] The internal standard curve method is referred to as the internal standard method. It calculates by measuring the relative values ​​of the peak areas of the internal standard and the measured component. It can eliminate the errors caused by changes in operating conditions to a certain extent, such as changes in injection volume, slight changes in chromatographic operating conditions, etc., thereby improving the accuracy of the analysis results.

[0087] The internal standard method can partially compensate for the loss of predicted components during sample pretreatment (such as concentration, extraction, derivatization, etc.) and reduce systematic errors. Among them, systematic error refers to the error caused by certain fixed reasons in scientific experiments or measurements. These errors always appear in the same way in repeated measurements and have predictable deviations. Systematic error is different from random error. Random error occurs by chance, and the results of each measurement may be different, while systematic error always appears in the same amount under the same conditions.

[0088] Therefore, the combination of GC-F ID and internal standard curve method can improve the accuracy and precision of quantitative analysis and reduce systematic errors.

[0089] S205. Obtain first flavor compound data based on the first category set and the first content set.

[0090] The first category set and the first content set are combined to obtain first flavor compound data. In the first flavor compound data, the category and content of the same flavor compound correspond one to one.

[0091] In one embodiment, based on LLME-GC-MS and internal standard curve method, the second flavor compound data of the liquor to be identified is obtained, including steps S301-S305:

[0092] S301, preparing a second sample to be tested based on the liquor to be identified.

[0093] For example, 10 mL of the liquor to be identified diluted to 10% vol was aspirated, sodium chloride was added to saturation, 10 μl of internal standard (2-octanol: 16.1 mg / L) was added, 2 ml of ether:pentane (1:1) was added, vortexed for 3 min, and after standing for stratification, the upper organic phase was aspirated into the injection bottle to obtain the second sample to be tested.

[0094] S302, using LLME-GC-MS to detect the second sample to be tested to obtain a second detection spectrum.

[0095] For example, GC (gas chromatography) conditions for LLME-GC-MS include:

[0096] The model of the chromatographic column is DB-FFAP, the specification of the chromatographic column is 60m×0.25mm×0.25μm, the brand of the chromatographic column is J&W Scientific, the injection port temperature and the gas chromatography detector temperature are both 250°C, helium is used as the carrier gas (purity ≥99.999%), the flow rate is 1.2ml / min, and non-split injection is used;

[0097] The heating program was as follows: maintain the initial temperature at 40°C for 1 min, increase the temperature to 100°C at a heating rate of 2.5°C / min, maintain for 1 min, increase the temperature to 160°C at a heating rate of 3°C / min, and then increase the temperature to 230°C at a heating rate of 5°C / min and maintain for 10 min.

[0098] The MS (mass spectrometry) conditions of LLME-GC-MS include:

[0099] An EI ion source and an electron energy of 70 eV were used, the ion source temperature was set to 230°C, the quadrupole temperature was set to 150°C, and the scanning range was m / z 30-550.

[0100] S303, determining a second category set of flavor compounds in the liquor to be identified according to the second detection spectrum.

[0101] For the specific process, please refer to the description of determining the first category set.

[0102] S304, using an internal standard curve method to quantify each flavor compound in the second category set to obtain a second content set of flavor compounds.

[0103] S305. Obtain second flavor compound data based on the second category set and the second content set.

[0104] The second type set and the second content set are combined to obtain second flavor compound data. In the second flavor compound data, the type and content of the same flavor compound correspond one to one.

[0105] In one embodiment, based on HPLC and external standard curve method, the third flavor compound data of the liquor to be identified is obtained, including steps S401-S405:

[0106] S401, preparing a third sample to be tested based on the liquor to be identified.

[0107] For example, 1 ml of the liquor to be identified is taken and passed through an organic filter membrane with a pore size of 0.22 μm to obtain a third sample to be tested.

[0108] S402, using HPLC to detect the third sample to be tested to obtain a third detection spectrum.

[0109] For example, first prepare the mobile phase. The specific preparation process includes: accurately weigh 2.72g KH2PO4 (potassium dihydrogen phosphate) in a beaker, dissolve it with an appropriate amount of ultrapure water and transfer it to a 1000ml volumetric flask, dilute to the scale with ultrapure water, adjust the pH to 2.3 with H3PO4, and filter with a 0.22μm water filter membrane.

[0110] Then perform HPLC detection, the conditions of HPLC include:

[0111] The model of the chromatographic column is SB-C18, the specifications of the chromatographic column are 4.6×150mm, 5-Micron (the particle diameter of the chromatographic column is 5 microns), the brand of the chromatographic column is ZORBAX, the mobile phase is 100% 0.02mol / L KH2PO4, pH=2.3, the detection wavelength is 208nm, the column temperature is 30°C, the mobile phase flow rate is 0.8ml / min, and the injection volume is 1μl.

[0112] S403, determining a third category set of flavor compounds in the liquor to be identified according to the third detection spectrum.

[0113] For the specific process, please refer to the description of determining the first category set.

[0114] S404, quantifying each flavor compound in the third category set using an external standard curve method to obtain a third content set of flavor compounds.

[0115] The external standard curve method is referred to as the external standard method. It is simple to operate, does not require complicated sample pretreatment, and does not require the addition of internal standards. Standard samples of known concentrations can be used directly to establish a standard curve to perform quantitative analysis on unknown samples.

[0116] The external standard curve method is simple to operate and more efficient, and can save a lot of time and internal standard costs. Therefore, on the basis of the two existing internal standard methods for quantitative analysis, which can ensure quantitative accuracy, the external standard method can be combined with HPLC to improve the detection efficiency.

[0117] S405. Obtain third flavor compound data based on the third category set and the third content set.

[0118] The third type set and the third content set are combined to obtain third flavor compound data. In the third flavor compound data, the type and content of the same flavor compound correspond one to one.

[0119] Finally, the union of the first category set, the second category set and the third category set can be obtained to obtain the target category set, and then the target content of the flavor compound can be determined according to the content of the same flavor compound in the first content set, the second content set and the third content set, and then the target content of each flavor compound in the target category set can be determined to obtain the target content set.

[0120] For example, for a flavor compound in a target category set, if the contents in the first content set, the second content set, and the third content set are different, the average of the contents of the flavor compound in the first content set, the second content set, and the third content set can be calculated as the target content.

[0121] Based on the target type set and the target content set, target flavor compound data is obtained. In the target flavor compound data, the types and contents of the flavor compounds correspond one to one.

[0122] In one embodiment, the physical and chemical index data includes at least one of total acid content, total ester content and solid content.

[0123] For example, the standard "Analysis Method for Liquor" (GB / T 10345-2022) can be referred to, and the indicator method can be used to determine the total acid content (measured as acetic acid content) and total ester content (measured as ethyl acetate content) in the liquor to be identified, and the drying method can be used to determine the solid content in the liquor to be identified.

[0124] In one embodiment, the mineral element data is obtained based on inductively coupled plasma mass spectrometry (ICP-MS) detection.

[0125] For example, referring to the National Food Safety Standard Determination of Multiple Elements in Food (GB-5009.268-2016), ICP-MS is used to determine the mineral elements in the liquor to be identified, and the specific mass number (mass-to-charge ratio, m / s) of the element is used for qualitative analysis. The intensity ratio of the mass spectrometry signal of the element to be determined and the mass spectrometry signal of the internal standard element is proportional to the concentration of the element to be determined for quantitative analysis. Among them, microwave digestion can be used for digestion of the liquor to be identified, and the other experimental conditions can be performed according to the standard.

[0126] S102: Determine the production area of ​​the liquor to be identified based on the first feature data set.

[0127] In one embodiment, step S102 includes: inputting the first feature data set into the liquor production area prediction model to obtain the predicted production area output by the liquor production area prediction model.

[0128] Among them, the liquor production area prediction model is trained based on the training data set and the machine learning model. The training data set includes a second feature data set of liquor samples from multiple production areas. The second feature data set includes flavor compound data samples, physical and chemical index data samples, and mineral element data samples. The label corresponding to the second feature data set includes the production area of ​​the liquor sample corresponding to the second feature data set.

[0129] Generally speaking, the training process of the liquor production area prediction model includes:

[0130] Obtain a second feature data set of liquor samples from multiple production areas;

[0131] Divide the second feature data set into a training data set and a test data set;

[0132] Perform model training based on the training data set to obtain the model to be verified;

[0133] The prediction accuracy of the model to be verified is verified according to the test data set until the prediction accuracy reaches the preset requirements, and the liquor production area prediction model is obtained.

[0134] However, most current models are trained under the condition of balanced sample distribution. However, in the actual liquor market, there is usually an imbalance in the number of liquors produced in different regions, and the commonly used classification methods often cannot fully learn the sample information of the minority class, making it difficult to achieve the ideal classification effect.

[0135] To this end, the model can be balanced before training, which can effectively reduce the model deviation caused by data imbalance and improve the performance of model classification.

[0136] Exemplarily, the training process of the liquor production area prediction model includes:

[0137] Obtain a second feature data set of liquor samples from multiple production areas;

[0138] Dividing the second feature data set into a first training data set and a test data set;

[0139] Performing equalization processing on the first training data set to obtain a second training data set;

[0140] Performing model training based on the second training data set to obtain a model to be verified;

[0141] The prediction accuracy of the model to be verified is verified according to the test data set until the prediction accuracy reaches the preset requirements, and the liquor production area prediction model is obtained.

[0142] In one embodiment, the training process of the liquor production area prediction model is as follows: Figure 2 As shown:

[0143] First, the flavor compound data, physical and chemical index data, and mineral element data of liquor samples from multiple production areas are obtained, and the flavor compound data, physical and chemical index data, and mineral element data corresponding to the same liquor sample are used as a set of feature data. The flavor compound data, physical and chemical index data, and mineral element data in each set of feature data are standardized and centralized, respectively, and merged to obtain a multidimensional data set. The multiple sets of feature data in the multidimensional data set are classified according to the production area to obtain a second feature data set of liquor samples from multiple production areas.

[0144] Then, the second feature data set is split into a first training data set and a test data set according to a preset ratio, wherein the preset ratio can be set and adjusted as needed.

[0145] The first training data set is equalized to obtain a new training data set, namely, the second training data set.

[0146] Exemplarily, the first training data set may be balanced using methods such as SMOTE oversampling, random oversampling, ensemble learning, and data enhancement.

[0147] Select the initial model for training, set the cross-validation parameters and grid search parameters, and use the quantitative data in the second feature data set as the independent variable. The quantitative data include the content of flavor compounds, total acid content, total ester content, solid content and the content of various mineral elements in the mineral element data. Take the production area as the dependent variable, specify the learning task and the activation function of the output layer, specify the number of classification categories according to the number of production areas, set the model parameters, and obtain the model to be trained.

[0148] Among them, machine learning models such as XGBoost model (extreme gradient boosting model), random forest model, logistic regression model and SVM (support vector machine) can be selected as initial models.

[0149] Finally, the training data set is input into the model to be trained to obtain the model to be verified, and the test data set is used to test the model to be verified to evaluate the performance of the model to be verified. If the performance is achieved, the training is stopped to obtain the liquor production area prediction model.

[0150] In order to improve the interpretability of the prediction results of the model, in one embodiment, after step S102, step S103 is further included:

[0151] S103. Calculate the contribution value of the first feature data set to the prediction result of the liquor production area prediction model to explain the prediction process of the liquor production area prediction model.

[0152] SHAP (SHapley Additive exPlanations) is a method for explaining the prediction results of machine learning models. It assigns a contribution value to each feature data input to the model, thereby explaining the model's prediction process.

[0153] Exemplarily, the SHAP method can be applied to calculate the contribution value of each feature data in the first feature data set of the liquor to be identified to the prediction result of the liquor production area prediction model. The contribution value can be divided into positive contribution and negative contribution according to its role, so as to further explain the relationship between each feature in the first feature data set and the predicted production area.

[0154] In addition, the contribution value of each feature data can be displayed in the form of a visual chart to facilitate user understanding.

[0155] Example 1

[0156] A total of 61 liquor samples were collected from the four major production areas in Guizhou Province, including 9 from Huichuan production area, 7 from Jinsha production area, 28 from Renhuai production area, and 17 from Xishui production area.

[0157] The flavor compounds in the liquor samples were qualitatively and quantitatively analyzed using GC-FID, LLME-GC-MS and HPLC to obtain flavor compound data samples. For details, please refer to the above process of using GC-FID, LLME-GC-MS and HPLC to obtain flavor compound data of liquor to be identified.

[0158] The total acid content, total ester content and solid content of the liquor sample are obtained as physical and chemical index data samples. For details, refer to the above process of obtaining the physical and chemical index data of the liquor to be identified.

[0159] ICP-MS is used to measure the mineral elements in the liquor sample to obtain the mineral element data sample of the liquor sample. For details, refer to the above process of obtaining the mineral element data of the liquor to be identified.

[0160] The flavor compound data samples, the physical and chemical index data samples and the mineral element data are used as the second characteristic data set of the liquor samples to obtain the second characteristic data sets of several liquor samples, and 80% of them are used as the training data set and 20% as the test data set.

[0161] The SMOTE oversampling method is used to equalize the training data set to obtain a new training data set.

[0162] Specifically, the training data set is input into the SmoteClassif() function in the UBL package (a software package in R language). The SMOTE parameters are set as C.perc = "balance", dist = "Euclidean".

[0163] Among them, C.perc is used to control the proportion of the second feature data sets of different production areas in the generated new training data set, C.perc = "balance" indicates that the proportion of the second feature data sets of different production areas in the generated new training data set is balanced, dist is used to specify the distance metric used to calculate the nearest neighbor, and dist = "Euclidean" indicates that the Euclidean distance is used to calculate the distance between the second feature data sets.

[0164] See also Figure 3 Before equalization, the training data set includes 7 sets of feature data from Huichuan, 5 sets of feature data from Jinsha, 22 sets of feature data from Renhuai, and 15 sets of feature data from Xishui. After equalization using SMOTE oversampling, the new training data set includes 12 sets of feature data from Huichuan, 12 sets of feature data from Jinsha, 12 sets of feature data from Renhuai, and 11 sets of feature data from Xishui.

[0165] Select the XGBoost model as the initial model for training, use the trainControl function to set the cross-validation parameters, use the expand.grid function to search the parameters, and then use the flavor compound data, physical and chemical index data, and mineral element data in the new training data set as independent variables, and the production area (Huichuan, Jinsha, Renhuai, Xishui) as the dependent variable to set the model parameters.

[0166] The specific model parameters are set as: objective = "multi:softmax", num_class = 4, nrounds = 20, max_depth = 2, eta = 0.2, gamma = 0.5, colsample_bytree = 0.6, min_child_weight = 3, subsample = 0.6.

[0167] Among them, objective = "multi:softmax" is used to specify that the learning task is a multi-classification task, and to specify the use of the softmax function as the activation function of the output layer; num_class is the number of classification categories; nrounds is the number of training iterations; max_depth is the maximum depth of the tree; eta is the learning rate; gamma is the minimum loss reduction during tree construction; colsample_bytree is used to control the proportion of randomly extracted features; min_child_weight is the minimum sum of weights required for a child object; subsample is used to control the proportion of random sampling.

[0168] Then use xgboost::xgb.DMatrix (a data structure dedicated to the XGBoost model, used to organize the input data into a specific format so that the XGBoost model can process it efficiently) to convert the new training data set into the format required for XGBoost modeling, and pass the format-converted training data set and the set model parameters to the xgb.train function to establish the model to be trained and train it to obtain the first model to be verified.

[0169] Finally, the test data set is input into the first model to be verified to verify the prediction accuracy of the first model to be verified, and the Confusion matrix is ​​used to calculate Accuracy, Precision, Recall, AUC (receiver operating characteristic curve), Kappa coefficient (chi-square correlation coefficient), F1 (harmonic mean of precision and recall) to comprehensively evaluate the performance of the first model to be verified. The evaluation results are shown in Table 1.

[0170] Table 1 Performance evaluation results of the first model to be verified

[0171] Accuracy Precision Recall F1 Kappa AUC 0.917 0.917 0.958 0.928 0.880 0.917

[0172] Assuming that the preset requirements for prediction accuracy include that the accuracy and precision are both greater than 0.8, according to the performance evaluation results in Table 1, it can be seen that the prediction accuracy of the first model to be verified has reached the preset requirements, and the training can be stopped. The first model to be verified can be used as the first liquor production area prediction model.

[0173] The first liquor production area prediction model can be used to predict the production areas of sauce-flavor liquor brewed in the four production areas of Huichuan, Jinsha, Renhuai and Xishui in Guizhou Province. The prediction results are shown in Table 2.

[0174] Table 2 Prediction results of Maotai-flavor liquors brewed in different production areas of Guizhou Province

[0175] Producing area Forecast quantity Forecast the correct quantity Number of forecast errors Huichuan 3 2 1 Sands 2 2 0 Renhuai 5 5 0 Xi Shui 2 2 0

[0176] The SHAP method is applied to calculate the contribution value of the feature data in the training data set to each prediction result (Huichuan production area, Jinsha production area, Renhuai production area and Xishui production area) of the first liquor production area prediction model, and generate the following Figure 4 The diagram shown. Figure 4 In the figure, the dark bar shows that the contribution value of a certain feature data to the prediction result of the first liquor producing area prediction model is a negative contribution, and the light bar shows that the contribution value of a certain feature data to the prediction result of the first liquor producing area prediction model is a positive contribution.

[0177] Among them, C66 (isobutyraldehyde), C51 (isobutanol) and C68 (isovaleraldehyde) make positive contributions to the Huichuan production area, while C50 (propanol), C38 (ethyl linoleate), C45 (β-phenylethanol) and Total.acid (total acid) make negative contributions to the Huichuan production area.

[0178] The contribution values ​​of C27 (ethyl acetate), C73 (valeric acid) and Mg (magnesium) to the Jinsha production area are positive, while the contribution values ​​of C32 (ethyl hexanoate), Total.ester (total ester), C22 (acetic acid), C81 (trimethylpyrazine) and C15 (ethyl phenylacetate) to the Jinsha production area are negative.

[0179] Cu (copper), C81 (trimethylpyrazine), C33 (propionic acid), Ca (calcium), C49 (sec-butyl alcohol), C36 (ethyl palmitate) and Co (cobalt) make positive contributions to the Renhuai production area, while C54 (isoamyl alcohol), C83 (2,3-diethylpyrazine) and C77 (2,5-dimethylpyrazine) make negative contributions to the Renhuai production area.

[0180] The contribution values ​​of V (vanadium), Total.acid (total acid), and C80 (2,3-dimethylpyrazine) to the Xishui production area are positive, while the contribution values ​​of C36 (ethyl palmitate), C38 (ethyl linoleate), Mn (manganese), Total.ester (total ester), C89 (4-ethylguaiacol) and C92 (dimethyl disulfide) to the Xishui production area are negative.

[0181] In this embodiment, the SMOTE oversampling method is used to process the training data set to increase the balance of the training data set, which can avoid the problem of decreased accuracy in producing area identification due to the imbalance in the number of samples between liquors from different producing areas.

[0182] In addition, a variety of evaluation indicators were used to jointly evaluate the performance of the model, among which Accuracy could reach above 0.91, Precision could reach above 0.82, Recall could reach above 0.82, F1 could reach above 0.82, Kappa coefficient could reach above 0.71, and AUG could reach above 0.91, indicating that the model can objectively and accurately predict the production area of ​​the liquor to be identified.

[0183] What is special is that the trained liquor production area identification model covers a data collection consisting of three dimensions: flavor components, physical and chemical indicators, and mineral elements. Compared with other prediction models in the same field, it can more comprehensively and systematically analyze the uniqueness of product quality in different production areas, enabling the model to achieve better prediction results in practical applications.

[0184] In addition, this embodiment also applies the SHAP method to measure the importance of different feature data in the first feature data set of the liquor to be identified, and obtains the contribution value of each feature data to the prediction result, so as to realize the interpretability of the prediction result of the XGBoost model, and at the same time provide a theoretical basis for the liquor production area prediction process.

[0185] In addition to Example 1, an attempt is also made to use a training data set that has not been equalized, and a training data set that has been equalized using the other two parameters HEOM and HVDM included in the SMOTE function for model training, and the performance of the trained model is evaluated. The evaluation results are shown in Table 3.

[0186] The SMOTE parameters can be set to C.perc="balance", dist="HEOM", or to C.perc="balance", dist="HVDM". dist="HEOM" indicates that the normalized Euclidean distance is used to calculate the distance between the second feature data sets, and dist="HVDM" indicates that the heterogeneous value difference metric distance is used to calculate the distance between the second feature data sets.

[0187] Table 3 Model performance evaluation results based on training data sets processed with different SMOTE parameters

[0188]

[0189] According to the results presented in Table 3, when the SMOTE parameters are set to C.perc = "balance" and dist = "Euclidean", the classification performance of the liquor production area prediction model trained based on the XGBoost model is better.

[0190] In addition to Example 1, attempts were made to use the balanced training data set in Example 1 to train SVM, random forest model and logistic regression model respectively, and the performance of the trained models was evaluated using the test data set in Example 1. The evaluation results are shown in Table 4.

[0191] Table 4 Performance evaluation results of different models

[0192]

[0193] According to the results presented in Table 4, the classification performance of the liquor production area prediction model trained based on the XGBoost model is good.

[0194] Example 2

[0195] A total of 88 liquor samples were collected from different producing areas on the market, including 61 from Guizhou, 2 from Guangxi, 5 from Hunan, 18 from Sichuan, and 2 from Shandong.

[0196] A training data set and a test data set of several liquor samples are obtained in the same manner as in Example 1.

[0197] The SMOTE oversampling method is used to equalize the training data set to obtain a new training data set.

[0198] Specifically, the training data set is input into the SmoteClassif() function in the UBL package (a software package in R language). The SMOTE parameters are set as C.perc = "balance", dist = "Euclidean".

[0199] See also Figure 5 Before equalization, the training data set includes 51 sets of feature data from Guizhou, 1 set of feature data from Guangxi, 5 sets of feature data from Hunan, 13 sets of feature data from Sichuan, and 1 set of feature data from Shandong. After equalization using SMOTE oversampling, the new training data set includes 14 sets of feature data from Guizhou, 13 sets of feature data from Guangxi, 13 sets of feature data from Hunan, 13 sets of feature data from Sichuan, and 13 sets of feature data from Shandong.

[0200] The XGBoost model was selected as the initial model for training, the cross-validation parameters were set using the trainControl function, the grid search parameters were set using the expand.grid function, and the flavor compound data, physical and chemical index data, and mineral element data in the new training data set were used as independent variables, and the production area (Guizhou, Guangxi, Hunan, Sichuan, and Shandong) was used as the dependent variable to set the model parameters.

[0201] The specific model parameters are set as: objective = "multi:softmax", num_class = 5, nrounds = 20, max_depth = 2, eta = 0.1, gamma = 0, colsample_bytree = 0.5, min_child_weight = 3, subsample = 1.

[0202] Similarly, xgboost::xgb.DMatrix is ​​used to convert the training data set into the format required for XGBoost modeling, and the converted training data set and the above parameters are passed to the xgb.train function to establish the model to be trained and train it to obtain the second model to be verified.

[0203] Finally, the test data set is input into the second model to be verified to verify the prediction accuracy of the second model to be verified, and the Confusion matrix is ​​used to calculate Accuracy, Precision, Recall, AUC, Kappa coefficient, and F1 to comprehensively evaluate the performance of the second model to be verified. The evaluation results are shown in Table 5.

[0204] Table 5 Performance evaluation results of the second model to be verified

[0205] Accuracy Precision Recall F1 Kappa AUC 0.929 0.826 0.826 0.826 0.711 0.925

[0206] Assuming that the preset requirements for prediction accuracy include that the accuracy and precision are both greater than 0.8, according to the performance evaluation results in Table 3, it can be seen that the prediction accuracy of the second model to be verified has reached the preset requirements, and the training can be stopped. The second model to be verified can be used as the second liquor production area prediction model.

[0207] The second liquor production area prediction model can be used to predict the production areas of the sauce-flavor liquor brewed in the five production areas of Guizhou, Guangxi, Hunan, Sichuan and Shandong. The prediction results are shown in Table 6.

[0208] Table 6 Prediction results of Maotai-flavor liquors brewed in different production areas

[0209] Producing area Forecast quantity Forecast the correct quantity Number of forecast errors Guizhou 10 9 1 Guangxi 1 1 0 Hunan 0 0 0 Sichuan 5 4 1 Shandong 1 0 1

[0210] The SHAP method is applied to calculate the contribution value of the feature data in the training data set to each prediction result of the first liquor production area prediction model (Guizhou production area, Guangxi production area, Hunan production area, Sichuan production area and Shandong production area), and the following is generated: Figure 6 The diagram shown. Figure 5 In the figure, the dark bar shows that the contribution value of a certain feature data to the prediction result of the first liquor producing area prediction model is a negative contribution, and the light bar shows that the contribution value of a certain feature data to the prediction result of the first liquor producing area prediction model is a positive contribution.

[0211] Example 3

[0212] Based on the flavor compound data, physical and chemical index data, and mineral element data obtained from the 61 liquor samples in Example 1, training data sets including different data types are formed respectively, and test data sets including the same data type are formed for each training data set. For example, the test data set 1 corresponding to the training data set 1 also only includes flavor compound data.

[0213] Training dataset 1 only includes flavor compound data of liquor samples;

[0214] Training data set 2 only includes the physical and chemical index data of liquor samples;

[0215] Training data set 3 only includes mineral element data of liquor samples;

[0216] Training dataset 4 only includes flavor compound data and physicochemical index data of liquor samples;

[0217] Training dataset 5 only includes flavor compound data and mineral element data of liquor samples;

[0218] Training data set 6 only includes the physical and chemical index data and mineral element data of liquor samples;

[0219] The training data set 7 only includes the flavor compound data, physical and chemical index data, and mineral element data of the liquor samples.

[0220] The training data sets 1 to 7 are used to train the models to be trained in the embodiment respectively, and the corresponding test data sets are used to input the trained models. Accuracy, Precision, Recall, AUC, Kappa coefficient, and F1 are calculated according to the prediction results output by the model to evaluate the performance of the trained models. The evaluation results are shown in Table 7.

[0221] Table 7. Model performance evaluation results based on training data sets 1 to 7

[0222] Training Dataset Accuracy Precision Recall F1 Kappa AUC 1 0.833 0.776 0.776 0.766 0.547 0.767 2 0.833 0.666 0.666 0.666 0.483 0.656 3 0.708 0.416 0.416 0.416 0.214 0.625 4 0.833 0.666 0.666 0.666 0.515 0.756 5 0.833 0.75 0.75 0.75 0.6 0.819 6 0.75 0.5 0.5 0.5 0.311 0.734 7 0.917 0.917 0.958 0.928 0.880 0.917

[0223] According to the results presented in Table 7, the performance evaluation result of the XGBoost model constructed based on training data set 7 is the best. That is, when the flavor compound data, physical and chemical index data, and mineral element data of the liquor to be identified are used as the first feature data set of the input liquor production area prediction model, the accuracy of predicting the production area is the highest.

[0224] It should be understood that although Figure 1 and 2The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1 and 2 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0225] Corresponding to the aforementioned method for identifying liquor producing areas, the present application also provides a device for identifying liquor producing areas.

[0226] In one embodiment, Figure 7 As shown, the identification device of the liquor producing area includes:

[0227] The data acquisition module 701 is used to acquire a first characteristic data set of the liquor to be identified, wherein the first characteristic data set includes flavor compound data, physical and chemical index data, and mineral element data;

[0228] The production area identification model 702 is used to determine the production area of ​​the liquor to be identified based on the first feature data set.

[0229] The specific definition of the identification device of the liquor production area can be found in the definition of the identification method of the liquor production area mentioned above, which will not be repeated here. Each module in the above-mentioned identification device of the liquor production area can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0230] The present application also provides a computer device. In one embodiment, the computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the steps of the method for identifying the liquor production area in the above embodiment are implemented.

[0231] In one embodiment, the computer device may be a server, and its internal structure diagram may be as follows: Figure 8As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store relevant data for implementing the above-mentioned identification method. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for identifying a liquor producing area is implemented.

[0232] In one embodiment, the computer device may be a terminal, and its internal structure diagram may be as follows: Fig. 9 As shown. The computer device includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for identifying a liquor production area is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a key, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0233] Those skilled in the art will understand that Figure 8 and Fig. 9 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0234] The present application also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the method for identifying a liquor producing area in the above embodiment are implemented.

[0235] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0236] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0237] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A method for identifying a liquor producing area, characterized in that: The identification method comprises: Acquire a first characteristic data set of liquor to be identified, wherein the first characteristic data set includes flavor compound data, physical and chemical index data, and mineral element data; The production area of ​​the liquor to be identified is determined based on the first feature data set.

2. The identification method according to claim 1, characterized in that: The flavor compound data include the types and contents of flavor compounds, and the types are detected based on at least one detection method of gas chromatography-flame ionization detector GC-FID, liquid-liquid microextraction-gas chromatography-mass spectrometry LLME-GC-MS and high performance liquid chromatography HPLC.

3. The identification method according to claim 2, characterized in that: Acquiring the flavor compound data of the liquor to be identified, including: Based on the GC-FID and internal standard curve method, obtaining the first flavor compound data of the liquor to be identified; Based on the LLME-GC-MS and internal standard curve method, obtaining the second flavor compound data of the liquor to be identified; Based on the HPLC and external standard curve method, obtaining the third flavor compound data of the liquor to be identified; The target flavor compound data of the liquor to be identified is obtained according to the first flavor compound data, the second flavor compound data and the third flavor compound data.

4. The identification method according to claim 2 or 3, characterized in that: The conditions of the GC-FID include: The model of the chromatographic column is DB-WAXUI, the specification of the chromatographic column is 30m×0.25mm×0.25μm, the injection port temperature and the hydrogen flame ionization detector temperature are both 250°C, the injection volume is 1μl, the split ratio is 30:1, the carrier gas is nitrogen (purity ≥99.999%), the column flow rate is 1ml / min, and the heating program is to maintain the initial temperature of 30°C for 3min, increase the temperature to 90°C at a heating rate of 3°C / min, and then increase the temperature to 180°C at a heating rate of 5°C / min, and maintain it for 25min; and / or, The gas chromatography GC conditions of the LLME-GC-MS include: The model of the chromatographic column is DB-FFAP, and the specification of the chromatographic column is 60m×0.25mm×0.25μm. The injection port temperature and the gas chromatography detector temperature are both 250°C. Helium is used as the carrier gas (purity ≥99.999%), the flow rate is 1.2ml / min, and the injection is non-split. The temperature program is to maintain the initial temperature of 40°C for 1min, increase the temperature to 100°C at a heating rate of 2.5°C / min, maintain it for 1min, increase the temperature to 160°C at a heating rate of 3°C / min, and then increase the temperature to 230°C at a heating rate of 5°C / min, and maintain it for 10min. The mass spectrometry MS conditions of the LLME-GC-MS include: using an EI ion source and an electron energy of 70 eV, setting the ion source temperature to 230° C., the quadrupole temperature to 150° C., and the scanning range to m / z 30 to 550; and / or, The HPLC conditions include: The model of the chromatographic column is SB-C18, the specifications of the chromatographic column are 4.6×150mm, 5-Micron, the mobile phase is 100% 0.02mol / L KH2PO4, pH=2.3, the detection wavelength is 208nm, the column oven is 30°C, the mobile phase flow rate is 0.8ml / min, and the injection volume is 1μl.

5. The identification method according to claim 1, characterized in that: The determining the production area of ​​the to-be-identified liquor based on the first feature data set includes: Inputting the first feature data set into a liquor production area prediction model to obtain a predicted production area output by the liquor production area prediction model; Among them, the liquor production area prediction model is trained based on a training data set and a machine learning model, the training data set includes a second feature data set of liquor samples from multiple production areas, the second feature data set includes flavor compound data samples, physical and chemical index data samples, and mineral element data samples, and the label corresponding to the second feature data set includes the production area of ​​the liquor sample corresponding to the second feature data set.

6. The identification method according to claim 5, characterized in that: After obtaining the predicted production area output by the liquor production area prediction model, the method further includes: The contribution value of the first feature data set to the prediction result of the liquor production area prediction model is calculated to explain the prediction process of the liquor production area prediction model.

7. The identification method according to claim 5, characterized in that: Before model training, the training data set is subjected to equalization processing; And / or, the liquor production area prediction model is obtained based on XGBoost model training.

8. The identification method according to claim 1, characterized in that: The physical and chemical index data include at least one of total acid content, total ester content and solid content; And / or, the mineral element data is obtained based on inductively coupled plasma mass spectrometer ICP-MS detection.

9. A device for identifying liquor producing areas, characterized in that: The identification device comprises: A data acquisition module, used to acquire a first characteristic data set of the liquor to be identified, wherein the first characteristic data set includes flavor compound data, physical and chemical index data, and mineral element data; A production area recognition model is used to determine the production area of ​​the liquor to be identified based on the first feature data set.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images