Distilled liquor classification method and device based on comprehensive two-dimensional gas chromatography

Abstract

The invention mainly relates to the technical field of wine brewing. The invention mainly provides a distilled liquor classification method and device based on comprehensive two-dimensional gas chromatography, and aims to quickly and accurately identify the category of distilled liquor. The core idea of the method comprises the following steps: acquiring comprehensive two-dimensional gas chromatograms of different types of distilled liquor samples, and preprocessing the comprehensive two-dimensional gas chromatograms to obtain comprehensive two-dimensional gas chromatography characteristics; constructing a distilled liquor classification model by taking a convolutional neural network as a basic model, constructing a distilled liquor classification data set according to the comprehensive two-dimensional gas chromatography characteristics and corresponding distilled liquor categories, and training the constructed distilled liquor classification model based on the constructed distilled liquor classification data set; the comprehensive two-dimensional gas chromatography characteristics of the distilled liquor sample to be detected are obtained and input into the distilled liquor classification model, the distilled liquor classification model outputs the category of the distilled liquor sample to be detected, and efficient and automatic classification of the distilled liquor sample can be achieved.

Description

Technical Field

[0001] The present invention mainly relates to the technical field of wine making, and in particular to a method and device for classifying distilled wine based on full two-dimensional gas chromatography. Background Art

[0002] With the rapid growth of the global distilled spirits market, consumers are gradually increasing their requirements for the quality and diversity of distilled spirits. Distilled spirits are diverse and of varying quality. Different brands, origins, years and other factors have a significant impact on the flavor and quality of the spirits. Traditional distilled spirit classification methods mainly rely on chemical composition analysis and manual experience, but this method is often complex, time-consuming and labor-intensive, and relies heavily on expert experience, making it difficult to achieve automated and efficient classification. Summary of the invention

[0003] Technical Problems to be Solved by the Invention

[0004] Provided are a distilled liquor classification method and device based on comprehensive two-dimensional gas chromatography, aiming to quickly and accurately identify the category of distilled liquor.

[0005] The technical solution adopted by the present invention to solve the above technical problems

[0006] In one aspect, the present invention provides a method for classifying distilled liquor based on comprehensive two-dimensional gas chromatography, comprising:

[0007] Obtaining full two-dimensional gas chromatograms of different types of distilled liquor samples, preprocessing the full two-dimensional gas chromatograms, and obtaining full two-dimensional gas chromatographic features;

[0008] A distilled liquor classification model is constructed based on a convolutional neural network as the basic model. A distilled liquor classification dataset is constructed according to the full two-dimensional gas chromatography characteristics and the corresponding distilled liquor categories. The constructed distilled liquor classification model is trained based on the constructed distilled liquor classification dataset.

[0009] The full two-dimensional gas chromatographic characteristics of the distilled liquor sample to be tested are obtained and input into the distilled liquor classification model, and the distilled liquor classification model outputs the category of the distilled liquor sample to be tested.

[0010] Furthermore, full two-dimensional gas chromatography analysis technology is used to obtain full two-dimensional gas chromatograms of different categories of distilled liquor samples according to the set one-dimensional time and two-dimensional time range, one-dimensional time and two-dimensional time drawing interval, color and numerical characterization standards.

[0011] Further, the preprocessing of the full two-dimensional gas chromatogram includes:

[0012] (1) Identification of key areas of full two-dimensional gas chromatograms: domain experts mark n chromatogram area blocks of key concern based on engineering experience, where n ≥ 1; each chromatogram area block is a rectangular area, and the rectangular area is located by the two diagonal points of the rectangular area; the location information of each chromatogram area block includes the one-dimensional time and the two-dimensional time corresponding to the two diagonal points of the rectangular area;

[0013] Alternatively, n chromatogram area blocks are directly obtained by obtaining a valid value area in the chromatogram and enclosing a rectangle;

[0014] (2) Comprehensive 2D gas chromatography feature reconstruction:

[0015] The n chromatogram area blocks are spliced ​​in the one-dimensional time direction and aligned downward, and the blank area is filled with the color representing 0 in the original chromatogram to obtain a rectangular chromatogram area block splicing feature; wherein, when it is 1, the chromatogram area block is directly used as the chromatogram area block splicing feature;

[0016] The chromatogram region block stitching features are resampled into an image of m×m pixel size; each channel of the image of m×m pixel size is normalized, and the data is normalized into a distribution with a mean of 0 and a standard deviation of 1 as a full two-dimensional gas chromatography feature.

[0017] Furthermore, the distilled liquor categories include the flavor, brand, region, year, type of liquor and category of grain used for brewing the distilled liquor.

[0018] Furthermore, the distilled liquor classification model includes a convolutional layer and a fully connected layer. The convolutional layer is used to extract pixel classification features in the full two-dimensional gas chromatogram features. The fully connected layer is an MLP network. The input feature dimension of the MLP network is a one-dimensional pixel classification feature. The output feature dimension of the MLP network is the number of distilled liquor categories.

[0019] Furthermore, the training of the constructed distilled liquor classification model includes:

[0020] The constructed distilled liquor classification dataset is divided into a training set and a validation set according to a set ratio;

[0021] The full two-dimensional gas chromatography features in the training set are used as input features of the convolutional neural network, and the corresponding distilled liquor category is used as the true output label of the convolutional neural network, and the true output label is represented by a one-hot vector;

[0022] During the training process, the predicted value of the distilled liquor category of the convolutional neural network model is compared with the corresponding true output label, the corresponding loss function is calculated, and the model is optimized through the back propagation of the convolutional neural network;

[0023] The distilled liquor classification model is tested using a validation set. When the loss function value of the validation set is lower than a preset loss threshold or the classification accuracy of the validation set is higher than a preset accuracy threshold, a trained distilled liquor classification model is obtained.

[0024] Furthermore, the loss function is: Where, loss is the loss function; k represents the kth distilled liquor sample; N is the total number of distilled liquor samples; C is the total number of distilled liquor categories; is the actual output category label of distilled liquor, where Indicates that the true category of distilled liquor is the i-th category, otherwise it is 0; is the predicted probability of the distilled liquor category, which represents the predicted probability of the i-th category of distilled liquor.

[0025] Furthermore, the method also includes: dynamically updating the distilled liquor classification data set, and automatically training and updating the distilled liquor classification model based on the updated distilled liquor classification data set.

[0026] On the other hand, the present invention also provides a distilled liquor classification device based on full two-dimensional gas chromatography, the device comprising a full two-dimensional gas chromatography feature acquisition module, a distilled liquor classification module, a distilled liquor classification model training module and a data dynamic update module;

[0027] The full two-dimensional gas chromatography feature acquisition module is used to obtain full two-dimensional gas chromatograms of different types of distilled liquor samples, and extract full two-dimensional gas chromatography features after preprocessing the full two-dimensional gas chromatograms;

[0028] The distilled liquor classification model training module is used to establish a dynamically updated distilled liquor classification data set, and automatically retrain the distilled liquor classification model based on the updated distilled liquor classification data set;

[0029] The distilled liquor classification module stores the distilled liquor classification model trained by the distilled liquor classification model training module, and is used to identify the type of the distilled liquor sample to be tested according to the full two-dimensional gas chromatographic characteristics of the distilled liquor sample to be tested;

[0030] The data dynamic update module is used to dynamically update the distilled liquor classification data set, and the distilled liquor classification model training module trains and updates the distilled liquor classification model based on the distilled liquor classification data set updated by the data dynamic update module.

[0031] Beneficial Effects of the Invention

[0032] The present invention applies convolutional neural networks to full two-dimensional gas chromatograms. Without the need for compound extraction, it can achieve efficient and automated classification of distilled liquor samples based directly on the full two-dimensional gas chromatogram data, thereby assisting experts in improving the accuracy and reliability of analysis. It is expected to enhance the level of intelligence in distilled liquor quality analysis, and assist engineers and experts in providing important support for the production, management and quality control of distilled liquor. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 The flowchart of the distilled liquor classification method based on comprehensive two-dimensional gas chromatography is as follows;

[0034] Figure 2 Diagram of the training process of the distilled spirits classification model;

[0035] Figure 3 Flow chart for preprocessing of full two-dimensional gas chromatograms;

[0036] Figure 4 This is a structural diagram of the distilled liquor classification model;

[0037] Figure 5 This is a schematic diagram of the distilled liquor classification device based on comprehensive two-dimensional gas chromatography; DETAILED DESCRIPTION

[0038] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present application more clear, the present application is further described in detail below in combination with the implementation methods. Unless otherwise defined, all scientific and technological terms used herein have the same meanings as understood by ordinary technicians in the field.

[0039] Obviously, the described embodiments are only some embodiments of the present invention, but not all 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.

[0040] In a first aspect, the distilled liquor classification method based on comprehensive two-dimensional gas chromatography of the present invention is as follows Figure 1 and Figure 2 As shown, including:

[0041] Step S1: obtaining full two-dimensional gas chromatograms of distilled liquor samples of different categories, preprocessing the full two-dimensional gas chromatograms, and obtaining full two-dimensional gas chromatogram features;

[0042] Step S2: constructing a distilled liquor classification model based on a convolutional neural network as a basic model, constructing a distilled liquor classification data set according to the full two-dimensional gas chromatography characteristics and the corresponding distilled liquor categories, and training the constructed distilled liquor classification model based on the constructed distilled liquor classification data set;

[0043] Step S3: Obtain the full two-dimensional gas chromatography characteristics of the distilled liquor sample to be tested, input the characteristics into the distilled liquor classification model, and the distilled liquor classification model outputs the category of the distilled liquor sample to be tested.

[0044] The step S1 of obtaining a comprehensive two-dimensional gas chromatogram of the distilled liquor sample comprises:

[0045] The full two-dimensional gas chromatogram is a two-dimensional color pixel image encoded in RGB color. The full two-dimensional gas chromatography analysis technology is used to obtain full two-dimensional gas chromatograms of different categories of distilled liquor samples in a manner that the full two-dimensional gas chromatogram of each sample needs to maintain the same one-dimensional time range, the same two-dimensional time range, the same one-dimensional time drawing interval, the same two-dimensional time drawing interval, and the same color and numerical characterization standards.

[0046] In this embodiment, the full two-dimensional gas chromatograms of all distilled liquor samples maintain the same one-dimensional time range [0s, 3000s] and the same two-dimensional time range [0s, 6s]; the same one-dimensional time and two-dimensional time drawing intervals are reflected by the same image resolution, which is 1584 pixels × 809 pixels. At the same time, when exporting the full two-dimensional gas chromatogram, the same color and numerical characterization standards are set.

[0047] The preprocessing process of the full two-dimensional gas chromatogram is as follows Figure 3 As shown, it includes comprehensive two-dimensional gas chromatography key area identification and comprehensive two-dimensional gas chromatography feature reconstruction;

[0048] (1) The identification of key areas of comprehensive two-dimensional gas chromatography includes the following two methods:

[0049] Domain experts recommend n chromatogram area blocks to focus on based on engineering experience, where n ≥ 1; each chromatogram area block is a rectangular area, and is positioned by two diagonal points of the rectangular area; the positioning information of each chromatogram area block includes one-dimensional time and two-dimensional time corresponding to the two diagonal points of the rectangular area; or it can be directly obtained by using a method of enclosing a valid numerical area in the chromatogram to obtain n chromatogram area blocks.

[0050] (2) Comprehensive 2D gas chromatography feature reconstruction includes two steps:

[0051] The n chromatogram area blocks are spliced ​​in the one-dimensional time direction and aligned downward, and the blank areas are filled with the color representing 0 in the original chromatogram to obtain a rectangular chromatogram area block splicing feature; when n is 1, the chromatogram area block is directly used as the chromatogram area block splicing feature.

[0052] The chromatogram region block stitching feature is resampled into an image of m×m pixel size: each channel of the m×m pixel size image is normalized, and the data is normalized to a distribution with a mean of 0 and a standard deviation of 1 as a full two-dimensional gas chromatography feature; the value of m includes one of the following: 32, 64, 128, 256, 512, 1024, and can also be determined by experienced engineers as other values; the normalization method is: x normalized =(x-mean) / std, where x normalized is the normalized value, mean is the mean of each channel of the image, and std is the standard deviation of each channel of the image.

[0053] In this embodiment, a rectangle is selected as the key area chromatogram area block, and the pixel coordinates corresponding to the diagonal points of the rectangle are (200, 50) and (1200, 600). Then, the area is reconstructed into a square image of 256×256 pixels using the resize function in the Python Imaging Library (PIL) library, and the resize method uses Image.LANCZOS;

[0054] In this embodiment, the normalization method is implemented using the transforms.Normalize function in the Torchvision library.

[0055] The distilled liquor classification data set is dynamically updated, and the distilled liquor classification model is automatically trained and updated based on the updated full two-dimensional gas chromatography classification data.

[0056] The classification standards of the distilled liquor category described in the present invention include: flavor, type, brand, region, year, and other classification standards that experts or enterprises are concerned about;

[0057] Specifically, the aroma types may include: strong aroma, sauce aroma, rice aroma, light aroma, soy aroma, Laobaigan aroma, mixed aroma, Dong aroma, Xiaoqu aroma, special aroma, sesame aroma, and Feng aroma; the aroma types may also include other aroma types that experts or enterprises need to classify;

[0058] Alcoholic beverages may include: brandy, whiskey, vodka, rum, gin, tequila, Chinese liquor; alcoholic beverages may also include other alcoholic beverages that experts or companies need to classify;

[0059] Brand type refers to the brands that experts or enterprises need to classify.

[0060] The regions may include: Yibin, Sichuan, Luzhou, Sichuan, Renhuai, Guizhou, Fenyang, Shanxi, and Cognac, France; the regions may also include other regions that experts or enterprises need to classify;

[0061] The years may include: 10 years, 20 years, 30 years, 50 years; the years may also include other years that experts or enterprises need to classify;

[0062] Other categories of interest to experts or businesses may also be included, for example, single and multiple grain categories for brewing grains.

[0063] In this embodiment, the predefined distilled liquor category feature is the single-grain and multi-grain category, including single-grain distilled liquor and multi-grain distilled liquor, which are represented by one-hot vectors, that is, the single-grain distilled liquor category feature is [1,0], and the multi-grain distilled liquor category feature is [0,1].

[0064] A distilled liquor classification model is constructed based on a convolutional neural network, and a distilled liquor classification dataset is constructed based on the full two-dimensional gas chromatographic features obtained in step S1 and the corresponding distilled liquor categories (represented by one-hot vectors) to train the constructed distilled liquor classification model:

[0065] Construction of distilled liquor classification model: Based on the convolutional neural network as the basic architecture, a distilled liquor classification model is constructed. Figure 4 As shown, it includes a convolutional neural network module and a fully connected module;

[0066] The convolutional neural network module includes any one of LeNet, AlexNet, ZFNet, VGG, GoogLeNet, ResNet, and DenseNet, and is used to extract image features in the full two-dimensional gas chromatogram;

[0067] The fully connected module is an MLP network, the input feature dimension of the MLP network is the output dimension of the one-dimensional convolutional neural network module, and the output feature dimension of the MLP network is the number of distilled liquor categories;

[0068] In this embodiment, the convolutional neural network module adopts the ResNet18 neural network, which is constructed by the torchvision.models.resnet18 function in the Torchvision library; the fully connected module adopts a one-layer Linear MLP network, which is constructed by the torch.nn.Linear function in the PyTorch library; the output feature dimension of the convolutional neural network module is 512, and the output feature dimension of the fully connected module is 2.

[0069] The distilled liquor classification data set is divided into a training set and a validation set according to a ratio, and the ratio of the training set to the validation set is usually 7:3, 8:2, or 9:1;

[0070] The full two-dimensional gas chromatography features in the training set are used as the input features of the convolutional neural network, and the corresponding distilled liquor category is used as the true output label of the convolutional neural network;

[0071] During the training process, the predicted value of the distilled liquor category of the convolutional neural network model is compared with the corresponding true output label, the corresponding loss function is calculated, and the model is optimized through the back propagation of the convolutional neural network; the loss function used in the calculation of the loss is the focal loss function: Where, loss is the loss function; k represents the kth distilled liquor sample; N is the total number of distilled liquor samples; C is the total number of distilled liquor categories; is the actual output category label of distilled liquor, where Indicates that the true category of distilled liquor is the i-th category, otherwise it is 0; is the predicted probability of the distilled liquor category, which represents the predicted probability of the i-th distilled liquor; where γ is a positive hyperparameter, usually 0, 1, 2 or 3. When it is 0, the focal loss function is the cross entropy loss function.

[0072] The distilled liquor classification model is tested using a validation set. When the loss function value of the validation set is lower than a preset loss threshold or the classification accuracy of the validation set is higher than a preset accuracy threshold, a trained distilled liquor classification model is obtained.

[0073] In this embodiment, the distilled liquor classification data set includes 20 single-grain distilled liquor data and 20 multi-grain distilled liquor data. The data set is divided into a training set and a validation set at an 8:2 ratio. Therefore, there are 4 single-grain distilled liquor data and 4 multi-grain distilled liquor data in the validation set.

[0074] The distilled liquor classification model is trained using a training set, and the distilled liquor classification model is tested using a validation set, and when the loss function value of the validation set is lower than a preset loss threshold or the classification accuracy of the validation set is higher than a preset accuracy threshold, a trained distilled liquor classification model is obtained;

[0075] In this embodiment, the accuracy of the validation set can reach 100%. In practical applications, the pre-set accuracy threshold is recommended to be no less than 95%, and the pre-set loss threshold is recommended to be no greater than 0.001. Taking into account the differences in data characteristics of different distilled spirits categories and the differences in the numerical ranges of different loss function calculation results, the pre-set loss threshold should be determined by engineers based on actual conditions.

[0076] When the distilled liquor classification dataset is updated, the distilled liquor classification model needs to be retrained using the updated distilled liquor classification dataset.

[0077] In a second aspect, the present invention provides a distilled liquor classification device based on comprehensive two-dimensional gas chromatography, such as Figure 5 As shown, it includes a full two-dimensional gas chromatography feature acquisition module, a distilled liquor classification module, a distilled liquor classification model training module and a data dynamic update module;

[0078] The full two-dimensional gas chromatography feature acquisition module is used to obtain full two-dimensional gas chromatograms of different types of distilled liquor samples, and extract full two-dimensional gas chromatography features after preprocessing the full two-dimensional gas chromatograms;

[0079] The distilled liquor classification model training module is used to establish a dynamically updated distilled liquor classification data set, and automatically retrain the distilled liquor classification model based on the updated distilled liquor classification data set;

[0080] The distilled liquor classification module stores the distilled liquor classification model trained by the distilled liquor classification model training module, and is used to identify the type of the distilled liquor sample to be tested according to the full two-dimensional gas chromatographic characteristics of the distilled liquor sample to be tested;

[0081] The data dynamic update module is used to dynamically update the distilled liquor classification data set, and the distilled liquor classification model training module trains and updates the distilled liquor classification model based on the distilled liquor classification data set updated by the data dynamic update module.

Claims

1. A method for classifying distilled liquor based on comprehensive two-dimensional gas chromatography, characterized in that: include: Obtaining full two-dimensional gas chromatograms of different types of distilled liquor samples, preprocessing the full two-dimensional gas chromatograms, and obtaining full two-dimensional gas chromatographic features; A distilled liquor classification model is constructed based on a convolutional neural network as the basic model. A distilled liquor classification dataset is constructed according to the full two-dimensional gas chromatography characteristics and the corresponding distilled liquor categories. The constructed distilled liquor classification model is trained based on the constructed distilled liquor classification dataset. The full two-dimensional gas chromatographic characteristics of the distilled liquor sample to be tested are obtained and input into the distilled liquor classification model, and the distilled liquor classification model outputs the category of the distilled liquor sample to be tested.

2. The method for classifying distilled liquors based on comprehensive two-dimensional gas chromatography according to claim 1, characterized in that: The method of obtaining the full two-dimensional gas chromatograms of different categories of distilled liquor samples includes: using the full two-dimensional gas chromatography analysis technology to obtain the full two-dimensional gas chromatograms of different categories of distilled liquor samples according to the set one-dimensional time and two-dimensional time range, one-dimensional time and two-dimensional time drawing interval, color and numerical characterization standards.

3. The method for classifying distilled liquor based on comprehensive two-dimensional gas chromatography according to claim 1, characterized in that: The preprocessing of the full two-dimensional gas chromatogram comprises: (1) Identification of key areas of full two-dimensional gas chromatograms: domain experts mark n chromatogram area blocks of key concern based on engineering experience, where n ≥ 1; each chromatogram area block is a rectangular area, and the rectangular area is located by the two diagonal points of the rectangular area; the location information of each chromatogram area block includes the one-dimensional time and the two-dimensional time corresponding to the two diagonal points of the rectangular area; Alternatively, n chromatogram area blocks are directly obtained by obtaining a valid value area in the chromatogram and enclosing a rectangle; (2) Comprehensive 2D gas chromatography feature reconstruction: The n chromatogram area blocks are spliced ​​in the one-dimensional time direction and aligned downward, and the blank area is filled with the color representing 0 in the original chromatogram to obtain a rectangular chromatogram area block splicing feature; wherein, when it is 1, the chromatogram area block is directly used as the chromatogram area block splicing feature; The chromatogram region block stitching features are resampled into an image of m×m pixel size; each channel of the image of m×m pixel size is normalized, and the data is normalized into a distribution with a mean of 0 and a standard deviation of 1 as a full two-dimensional gas chromatography feature.

4. The method for classifying distilled liquor based on comprehensive two-dimensional gas chromatography according to claim 1, characterized in that: The categories of distilled liquor include the flavor, brand, region, year, type of liquor and category of grain used for brewing.

5. The distilled liquor classification method based on comprehensive two-dimensional gas chromatography according to any one of claims 1 to 4, characterized in that: The distilled liquor classification model includes a convolutional layer and a fully connected layer. The convolutional layer is used to extract pixel classification features in the full two-dimensional gas chromatogram features. The fully connected layer is an MLP network. The input feature dimension of the MLP network is a one-dimensional pixel classification feature. The output feature dimension of the MLP network is the number of distilled liquor categories.

6. The method for classifying distilled liquor based on comprehensive two-dimensional gas chromatography according to claim 5, characterized in that: The training of the constructed distilled liquor classification model includes: The constructed distilled liquor classification dataset is divided into a training set and a validation set according to a set ratio; The full two-dimensional gas chromatography features in the training set are used as the input features of the convolutional neural network, and the corresponding distilled liquor category is used as the true output label of the convolutional neural network; During the training process, the predicted value of the distilled liquor category of the convolutional neural network model is compared with the corresponding true output label, the corresponding loss function is calculated, and the model is optimized through the back propagation of the convolutional neural network; The distilled liquor classification model is tested using a validation set. When the loss function value of the validation set is lower than a preset loss threshold or the classification accuracy of the validation set is higher than a preset accuracy threshold, a trained distilled liquor classification model is obtained.

7. The method for classifying distilled liquor based on comprehensive two-dimensional gas chromatography according to claim 6, characterized in that: The loss function is: Where, loss is the loss function; k represents the kth distilled liquor sample; N is the total number of distilled liquor samples; C is the total number of distilled liquor categories; is the actual output category label of distilled liquor, where Indicates that the true category of distilled liquor is the i-th category, otherwise it is 0; is the predicted probability of the distilled liquor category, which represents the predicted probability of the i-th category of distilled liquor.

8. The method for classifying distilled liquor based on comprehensive two-dimensional gas chromatography according to claim 6, characterized in that: The true output label is represented by a one-hot vector.

9. The method for classifying distilled liquor based on comprehensive two-dimensional gas chromatography according to claim 6, characterized in that: The training of the constructed distilled liquor classification model also includes: dynamically updating the distilled liquor classification data set, and the distilled liquor classification model is automatically trained and updated based on the updated distilled liquor classification data set.

10. A distilled liquor classification device based on comprehensive two-dimensional gas chromatography, used to implement the distilled liquor classification method based on comprehensive two-dimensional gas chromatography as described in any one of claims 1 to 9, characterized in that: It includes a full two-dimensional gas chromatography feature acquisition module, a distilled liquor classification module, a distilled liquor classification model training module and a data dynamic update module; The full two-dimensional gas chromatography feature acquisition module is used to obtain full two-dimensional gas chromatograms of different types of distilled liquor samples, and extract full two-dimensional gas chromatography features after preprocessing the full two-dimensional gas chromatograms; The distilled liquor classification model training module is used to establish a dynamically updated distilled liquor classification data set, and automatically retrain the distilled liquor classification model based on the updated distilled liquor classification data set; The distilled liquor classification module stores the distilled liquor classification model trained by the distilled liquor classification model training module, and is used to identify the type of the distilled liquor sample to be tested according to the full two-dimensional gas chromatographic characteristics of the distilled liquor sample to be tested; The data dynamic update module is used to dynamically update the distilled liquor classification data set, and the distilled liquor classification model training module trains and updates the distilled liquor classification model based on the distilled liquor classification data set updated by the data dynamic update module.

Images