Intelligent Baijiu blending system and method based on machine learning

Abstract

The invention provides an intelligent Baijiu blending system and method based on machine learning, and the method comprises the steps: carrying out the initial prediction of the blending ratio of Baijiu to base liquor through the flavor component contents of the Baijiu and the base liquor, and obtaining a first blending parameter; projection clustering is carried out on the near infrared spectrum data of the white spirit and the base liquor to construct a spectrum characteristic space, and then the baseline offset of each spectrum characteristic wavelength in the spectrum characteristic space is determined; further screening out the characteristic wavelengths of the white spirit and the base spirit from the spectral characteristic space through the baseline offset, and carrying out secondary prediction on the blending ratio of the white spirit to the base spirit according to the characteristic wavelengths to obtain a second blending parameter; and based on the first blending parameter and the second blending parameter, performing coupling training on the intelligent blending model, and further predicting the blending ratio between the white spirit and the base liquor through the trained intelligent blending model. By the adoption of the scheme, intelligent coupling estimation of the Baijiu blending ratio can be achieved, and therefore the blending precision in Baijiu blending is improved.

Description

Technical Field

[0001] The present application relates to the field of machine learning technology, and more specifically, to a white wine intelligent blending system and method based on machine learning. Background Art

[0002] Machine learning plays a key role in optimizing formulas and improving product consistency in the liquor blending process. By analyzing historical data and various parameters in the brewing process, machine learning models can identify key factors that affect liquor quality and predict the optimal combination of different raw materials. This technology not only improves blending efficiency, but also can quickly respond to changes in consumer taste preferences and flexibly adjust production strategies to ensure the high quality and market competitiveness of the final product.

[0003] In the existing technology, traditional liquor blending technology usually relies on a single data source, such as chemical composition (flavor substances) or sensory evaluation. The single data source method often cannot fully characterize the complexity of liquor and ignores the interaction between flavor components and physical properties, resulting in insufficient prediction accuracy of the blending ratio. The multi-dimensional data characteristics of liquor are coupled and analyzed using a machine learning model, which effectively integrates flavor components and near-infrared spectral data to capture the deep relationship between them. It can integrate chemical and physical information and make up for the one-sidedness caused by a single data source. Therefore, how to achieve intelligent coupled estimation of liquor blending ratio and thus improve the blending accuracy in liquor blending has become a difficult problem faced by the industry. Summary of the invention

[0004] The present application provides a system and method for intelligent liquor blending based on machine learning, which can realize intelligent coupled estimation of liquor blending ratio, thereby improving the blending accuracy in liquor blending.

[0005] In a first aspect, the present application provides a method for training a liquor intelligent blending model based on machine learning, which is used in a liquor intelligent blending system for performing liquor intelligent blending, comprising the following steps:

[0006] Construct an intelligent blending model for liquor blending based on twin neural network initialization;

[0007] Collecting the flavor component contents of the liquor and the base wine, and making an initial prediction of the blending ratio between the liquor and the base wine according to the flavor component contents of the liquor and the base wine, so as to obtain a first blending parameter of the intelligent blending model;

[0008] Collecting near-infrared spectral data of the liquor and base wine, performing projection clustering on the near-infrared spectral data to construct a spectral feature space, and then determining the baseline offset of each spectral feature wavelength according to the difference gradient between each spectral feature wavelength in the spectral feature space;

[0009] The characteristic wavelengths of the liquor and the base wine are screened out from the spectral feature space by the baseline offset of each spectral characteristic wavelength, and then the blending ratio between the liquor and the base wine is secondary predicted based on the screened characteristic wavelengths to obtain the second blending parameter of the intelligent blending model;

[0010] Based on the first blending parameter and the second blending parameter, the intelligent blending model is coupled trained, and then the blending ratio between the liquor and the base liquor is predicted by the trained intelligent blending model.

[0011] Preferably, the blending ratio between the liquor and the base wine is initially predicted according to the flavor component contents of the liquor and the base wine, and the first blending parameter of the intelligent blending model is obtained, which specifically includes:

[0012] Standardizing the flavor component contents of the liquor and base liquor to obtain standard flavor component content data;

[0013] The initial blending ratio of the liquor and the base liquor is predicted according to the standard flavor component content data, and the initial blending ratio is used as the first blending parameter of the intelligent blending model.

[0014] Preferably, performing projection clustering on the near-infrared spectral data to construct a spectral feature space specifically includes:

[0015] Extracting the main features of the near-infrared spectral data using principal component analysis to obtain a spectral feature set;

[0016] Performing band hierarchical clustering on the spectral features in the spectral feature set to obtain spectral feature clusters of multiple bands;

[0017] All spectral feature clusters are projected and mapped into a unified feature space to obtain a spectral feature space.

[0018] Preferably, determining the baseline offset of each spectral feature wavelength according to the difference gradient between each spectral feature wavelength in the spectral feature space specifically includes:

[0019] Determine the difference gradient between the wavelengths of each spectral feature in the spectral feature space using a difference method;

[0020] Based on a normal distribution assumption, determining a baseline wavelength of the spectral feature in the spectral feature space according to all difference gradients;

[0021] The baseline offset of the corresponding spectral characteristic wavelength is determined by the baseline wavelength and each spectral characteristic wavelength.

[0022] Preferably, screening out the characteristic wavelengths of the liquor and base wine from the spectral feature space by using the baseline offset of each spectral characteristic wavelength specifically includes:

[0023] Performing offset threshold comparison on the baseline offset of each spectral feature wavelength;

[0024] When the baseline offset is less than or equal to a preset offset threshold, the spectral characteristic wavelength corresponding to the baseline offset is taken as the effective spectral characteristic wavelength, and then all effective spectral characteristic wavelengths are obtained;

[0025] All effective spectral characteristic wavelengths are used as characteristic wavelengths of the liquor and base wine.

[0026] Preferably, based on the first blending parameter and the second blending parameter, coupling training of the intelligent blending model specifically includes:

[0027] Determining a coupling correlation between the first blending parameter and the second blending parameter;

[0028] Determining a sharing weight between two branch networks in the intelligent blending model according to the coupling correlation;

[0029] The first blending parameter and the second blending parameter are respectively used as prior parameters of two branch networks in the intelligent blending model, and then the intelligent blending model is trained using a preset training data set.

[0030] Preferably, the flavor component content refers to the specific concentration of various volatile compounds in the liquor and the base wine.

[0031] In a second aspect, the present application provides a liquor intelligent blending system based on machine learning, including a liquor intelligent blending model training unit, wherein the liquor intelligent blending model training unit includes:

[0032] A construction module is used to construct an intelligent blending model for liquor blending based on the twin neural network initialization;

[0033] A processing module, for collecting the flavor component contents of the liquor and the base wine, and making an initial prediction of the blending ratio between the liquor and the base wine according to the flavor component contents of the liquor and the base wine, so as to obtain a first blending parameter of the intelligent blending model;

[0034] The processing module is further used to collect near-infrared spectral data of the liquor and base wine, perform projection clustering on the near-infrared spectral data to construct a spectral feature space, and then determine the baseline offset of each spectral feature wavelength according to the difference gradient between each spectral feature wavelength in the spectral feature space;

[0035] The processing module is further used to filter out the characteristic wavelengths of the liquor and the base wine from the spectral feature space by using the baseline offset of each spectral characteristic wavelength, and then perform a secondary prediction on the blending ratio between the liquor and the base wine based on the filtered characteristic wavelengths to obtain a second blending parameter of the intelligent blending model;

[0036] An execution module is used to perform coupling training on the intelligent blending model based on the first blending parameter and the second blending parameter, and then predict the blending ratio between the liquor and the base liquor through the trained intelligent blending model.

[0037] In a third aspect, the present application provides a computer device, comprising a memory and a processor, wherein the memory stores codes, and the processor is configured to obtain the codes and execute the above-mentioned machine learning-based intelligent liquor blending model training method.

[0038] In a fourth aspect, the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the above-mentioned machine learning-based intelligent liquor blending model training method.

[0039] The technical solution provided by the embodiments disclosed in this application has the following beneficial effects:

[0040] In an embodiment of the present application, first, an intelligent blending model for liquor blending is constructed based on the initialization of a twin neural network; the flavor component contents of the liquor and the base wine are collected, and the blending ratio between the liquor and the base wine is initially predicted according to the flavor component contents of the liquor and the base wine, so as to obtain a first blending parameter of the intelligent blending model; near-infrared spectral data of the liquor and the base wine are collected, and a spectral feature space is constructed by projecting and clustering the near-infrared spectral data, and then the baseline offset of each spectral feature wavelength is determined according to the difference gradient between each spectral feature wavelength in the spectral feature space; the characteristic wavelengths of the liquor and the base wine are screened out from the spectral feature space according to the baseline offset of each spectral feature wavelength, and then the blending ratio between the liquor and the base wine is secondary predicted according to the screened characteristic wavelengths, so as to obtain a second blending parameter of the intelligent blending model; based on the first blending parameter and the second blending parameter, the intelligent blending model is coupled trained, and then the blending ratio between the liquor and the base wine is predicted by the trained intelligent blending model.

[0041] It can be seen that in this application, the intelligent blending model is coupled and trained based on the first blending parameter and the second blending parameter, and then the blending ratio between the liquor and the base wine is predicted by the trained intelligent blending model; wherein, firstly, the blending ratio between the liquor and the base wine is initially predicted according to the content of the flavor components, and the first blending ratio is obtained, and the first blending ratio is used as the first blending parameter of the intelligent blending model, and the first blending parameter acts as a priori knowledge in the intelligent blending model, so that the intelligent blending model can make reasonable blending ratio predictions in the initial stage, which can lay the foundation for the subsequent coupling optimization based on spectral data, enhance the processing ability of the intelligent blending model for multidimensional features, thereby improving the initial coupling effect between the flavor components and the near-infrared spectrum; secondly, based on the baseline offset of different wavelengths in the spectral feature space, the flavor components that best reflect the difference between the liquor and the base wine are selected. The spectral characteristic wavelength reflects the subtle changes in the physical properties of the liquor through the spectral characteristic wavelength, which can better supplement the lack of flavor components. On this basis, the second blending ratio is predicted using the screened spectral characteristic wavelength, and the second blending ratio is used as the second blending parameter of the intelligent blending model. Further, by performing correlation coupling analysis on the first blending parameter and the second blending parameter, the complementarity between different data sources (i.e., flavor components and near-infrared spectra) is achieved, which makes up for the one-sidedness caused by a single data source, and then facilitates the subsequent coupling training of the intelligent blending model to improve the ability to grasp the complex chemical (i.e., flavor components) and physical (i.e., near-infrared spectra) relationship in liquor blending, thereby improving the blending accuracy in liquor blending; in summary, the implementation plan of the present application can realize the intelligent coupling estimation of the liquor blending ratio, thereby improving the blending accuracy in liquor blending. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 is an exemplary flow chart of a liquor intelligent blending model training method based on machine learning in some embodiments of the present application;

[0043] Figure 2 is a schematic diagram of the structure of a twin neural network in some embodiments of the present application;

[0044] Figure 3 is a schematic diagram of a process for determining a spectral feature space in some embodiments of the present application;

[0045] Figure 4 It is a structural schematic diagram of a liquor intelligent blending model training unit in some embodiments of the present application;

[0046] Figure 5 It is a structural diagram of a computer device for implementing a white wine intelligent blending model training method based on machine learning in some embodiments of the present application. DETAILED DESCRIPTION

[0047] In order to better understand the technical solution of the present application, the technical solution of the application will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0048] refer to Figure 1 , which is an exemplary flow chart of a liquor intelligent blending model training method based on machine learning in some embodiments of the present application. The liquor intelligent blending model training method 100 based on machine learning mainly includes the following steps:

[0049] In step 101, an intelligent blending model for liquor blending is constructed based on the twin neural network initialization.

[0050] It should be noted that a Siamese Neural Network (SNN) is usually composed of two or more branch networks that share weights, and each branch network is responsible for processing different input features. In this application, an intelligent blending model for liquor blending is constructed based on the initialization of the twin neural network. The core is to use the structure of the twin neural network to compare and learn different input features to predict the optimal blending ratio between the liquor and the base wine.

[0051] For specific implementation, refer to Figure 2 This figure is a schematic diagram of the structure of the twin neural network in some embodiments of the present application. A dual-branch network structure can be designed first, and each branch network receives different characteristic information of the liquor and the base liquor, such as the flavor component content and the near-infrared spectrum data; then the flavor component content is used as the initialization parameter of the first branch network, and the near-infrared spectrum data is used as the initialization parameter of the second branch network; finally, the loss function of the twin neural network is designed, and the similarity is measured by the loss function. In this application, the contrast loss function (Contrastive Loss) can be used as the loss function of the twin neural network to complete the initialization construction of the intelligent blending model.

[0052] In step 102, the flavor component contents of the liquor and the base wine are collected, and an initial prediction is made on the blending ratio between the liquor and the base wine based on the flavor component contents of the liquor and the base wine to obtain a first blending parameter of the intelligent blending model.

[0053] It should be noted that the flavor component content described in the present application refers to the specific concentration of various volatile compounds in the liquor and base liquor, including alcohols, esters, acids, aldehydes, ketones and phenolic compounds. These components are quantitatively determined in gas chromatography-mass spectrometry analysis, which can provide basic data for the prediction of the blending ratio; because the flavor component content of different batches of liquor and base liquor may vary, the liquor and base liquor used in the training of this application are from the same batch as the liquor and base liquor used for blending the liquor after the training, and each batch undergoes a new round of training and blending.

[0054] In specific implementation, the collection of the flavor component content of liquor and base wine can be achieved in the following manner, namely: take an appropriate amount of liquor and base wine, first dilute and filter them to remove solid impurities, ensure that the sample is suitable for gas chromatography-mass spectrometry analysis, then select a suitable gas chromatography column (such as a capillary column), set a suitable temperature program, carrier gas flow rate and injection volume, separate the compounds, and then introduce the separated compounds into a mass spectrometer, set the ion source and mass spectrometry analysis conditions, select a suitable ionization method (such as electron bombardment or chemical ionization) for mass spectrometry detection, and then obtain a mass spectrum of each component through mass spectrometry analysis, use professional software for data processing, quantify the flavor components in liquor and base wine, and generate a flavor component spectrum, finally identify and quantify the main flavor components based on the generated flavor component spectrum, and use the content of the main flavor components as the flavor component content for subsequent blending ratio prediction.

[0055] In some embodiments, the blending ratio between the liquor and the base wine is initially predicted according to the flavor component content of the liquor and the base wine, and the first blending parameter of the intelligent blending model is obtained by the following steps:

[0056] Standardizing the flavor component contents of the liquor and base liquor to obtain standard flavor component content data;

[0057] The initial blending ratio of the liquor and the base liquor is predicted according to the standard flavor component content data, and the initial blending ratio is used as the first blending parameter of the intelligent blending model.

[0058] In specific implementation, first, the flavor component contents of the liquor and the base wine can be standardized by quantile standardization in the prior art, and then the data set obtained after the standardization is used as standard flavor component content data; then, the standard flavor component content data is used as the input parameter of the pre-trained liquor blending model, and the liquor blending model will analyze the features and relationships learned during training, and predict the blending ratio of the liquor and the base wine through the liquor blending model to obtain the initial blending ratio of the liquor and the base wine, and then use the initial blending ratio as the first blending parameter of the intelligent blending model; it should be noted that the first blending parameter in the present application refers to the blending ratio predicted based on the flavor component contents of the liquor and the base wine, which can be used as prior knowledge of the intelligent blending model.

[0059] It should be noted that the pre-trained liquor blending model in the present application refers to a machine learning model trained on historical flavor component content data, which aims to predict the blending ratio between liquor and base liquor. The liquor blending model uses known flavor component content data and corresponding blending ratios to learn the relationship and pattern between them, thereby realizing the prediction of the blending ratio; it should also be noted that the algorithm structure of the liquor blending model in the present application can adopt a support vector machine. In other embodiments, the algorithm structure of the liquor blending model can also adopt other algorithm structures, which are not limited here.

[0060] In step 103, near-infrared spectral data of the liquor and base wine are collected, and the near-infrared spectral data are projected and clustered to construct a spectral feature space, and then the baseline offset of each spectral feature wavelength is determined according to the difference gradient between each spectral feature wavelength in the spectral feature space.

[0061] It should be noted that the present application scheme is based on the different responses of near-infrared spectroscopy to the molecular vibration and rotation characteristics of chemical components (i.e., the chemical components of liquor and base wine), thereby quantitatively analyzing the component content in the sample. The principle of using near-infrared spectroscopy to determine the component content in liquor and base wine is as follows, namely: near-infrared spectroscopy is mainly analyzed through the vibration and rotation absorption characteristics of molecules. Different compounds (such as alcohols, esters, acids, etc.) have unique absorption peaks within a specific wavelength range; when near-infrared light is irradiated on the sample, the molecules in the sample will absorb light of a specific wavelength to produce an absorption spectrum. Each chemical component corresponds to a specific absorption band, so it can be identified and quantified through spectral characteristics; further, by establishing a standard curve or regression model of known components, the measured spectral data is compared with the component concentration of the known sample, thereby quantitatively analyzing the component content in the sample (liquor and base wine).

[0062] In specific implementation, the collection of near-infrared spectral data of the liquor and base wine can be achieved in the following manner, namely: first, take an appropriate amount of liquor and base wine samples, ensure that the samples are clean and free of bubbles, so as not to affect the spectral measurement results; then, set the appropriate wavelength range (usually 800-2500nm) according to the instrument manual, select the appropriate scanning speed and number, and then place the sample in the sample compartment of the spectrometer, start the instrument to collect the spectrum, and each sample can be measured multiple times to improve the reliability of the data; finally, the collected near-infrared spectral data is recorded in real time, usually stored in the form of a relationship diagram between spectral intensity and wavelength, thereby obtaining the near-infrared spectral data of the liquor and base wine.

[0063] In some embodiments, reference Figure 3 As shown, this figure is a schematic diagram of the process of determining the spectral feature space in some embodiments of the present application. In this embodiment, the projection clustering of the near-infrared spectral data to construct the spectral feature space can be implemented by the following steps:

[0064] In step 1031, principal component analysis is used to extract the main features of the near infrared spectrum data to obtain a spectrum feature set;

[0065] In step 1032, the spectral features in the spectral feature set are subjected to band hierarchical clustering to obtain spectral feature clusters of multiple bands;

[0066] In step 1033, all spectral feature clusters are projected and mapped to a unified feature space to obtain a spectral feature space.

[0067] In the specific implementation, first, all near-infrared spectral data are standardized, and then the standardized data are integrated into a matrix, in which the rows represent the compound components in the sample and the columns represent the wavelengths. The principal component analysis algorithm is further used to calculate the covariance matrix, and the first few principal components are extracted (usually the components with a cumulative contribution rate of 95% are selected), and the spectral features of all the extracted components are combined into a set as a spectral feature set; then, a hierarchical clustering algorithm is used to calculate the distance matrix based on the similarity between the spectral features in the spectral feature set (such as Euclidean distance or cosine similarity), and then similar spectral features are gradually merged according to the set distance threshold to form spectral feature clusters of multiple bands; finally, the principal components can be used as coordinate axes to construct a multidimensional feature space, and then the center point of each spectral feature cluster is mapped to the multidimensional feature space to form a spectral feature space.

[0068] In some embodiments, determining the baseline offset of each spectral feature wavelength according to the difference gradient between each spectral feature wavelength in the spectral feature space can be achieved by using the following steps:

[0069] Determine the difference gradient between the wavelengths of each spectral feature in the spectral feature space using a difference method;

[0070] Based on a normal distribution assumption, determining a baseline wavelength of the spectral feature in the spectral feature space according to all difference gradients;

[0071] The baseline offset of the corresponding spectral characteristic wavelength is determined by the baseline wavelength and each spectral characteristic wavelength.

[0072] In the specific implementation, first, the difference method in the prior art is used to calculate the average difference in wavelength between the selected spectral feature and other spectral features in the spectral feature space, and the average difference is used as the difference gradient of the wavelength between the selected spectral feature and other spectral features, so as to obtain the difference gradient of the wavelength between each spectral feature and other spectral features; assuming that each difference gradient conforms to the normal distribution, the average value of all difference gradients is calculated, and the average value is used as the gradient mean, and then the wavelength corresponding to or closest to the gradient mean is selected as the baseline wavelength. Assuming that the gradient mean corresponds to or closest to a wavelength of 600nm, the wavelength of 600nm is selected as the baseline wavelength; the wavelength of each spectral feature is obtained from the spectral feature space, and the Euclidean distance between the baseline wavelength and the wavelength of each spectral feature can be used as the baseline offset of the corresponding spectral feature wavelength.

[0073] It should be noted that the baseline offset in the present application is a measure of the degree of deviation of the spectral characteristic wavelength relative to its reference baseline. The larger the baseline offset, the greater the degree of deviation of the spectral characteristic wavelength relative to its reference baseline, and the smaller the baseline offset, the smaller the degree of deviation of the spectral characteristic wavelength relative to its reference baseline. It should also be noted that in the blending of liquor, the spectral data may be affected by background light, impurities in the sample or environmental factors, causing the spectral characteristics to shift up and down. By calculating the baseline offset, these interferences can be eliminated to obtain more accurate spectral information.

[0074] In step 104, the characteristic wavelengths of the liquor and the base wine are screened out from the spectral characteristic space through the baseline offset of each spectral characteristic wavelength, and then the blending ratio between the liquor and the base wine is secondary predicted based on the screened characteristic wavelengths to obtain the second blending parameter of the intelligent blending model.

[0075] In some embodiments, the following steps may be used to screen out the characteristic wavelengths of the liquor and base wine from the spectral feature space by using the baseline offset of each spectral characteristic wavelength:

[0076] Performing offset threshold comparison on the baseline offset of each spectral feature wavelength;

[0077] When the baseline offset is less than or equal to a preset offset threshold, the spectral characteristic wavelength corresponding to the baseline offset is taken as the effective spectral characteristic wavelength, and then all effective spectral characteristic wavelengths are obtained;

[0078] All effective spectral characteristic wavelengths are used as characteristic wavelengths of the liquor and base wine.

[0079] In specific implementation, first, an offset threshold can be set in advance, and the offset threshold can be set by a large number of baseline offsets. In some embodiments, the average value of all baseline offsets can be used as the offset threshold. In other embodiments, other methods can be used to set it, which are not limited here; secondly, the baseline offset of each spectral characteristic wavelength is compared with the offset threshold, and the spectral characteristic wavelength corresponding to the baseline offset less than or equal to the offset threshold is taken as the valid spectral characteristic wavelength, and the spectral characteristic wavelength corresponding to the baseline offset greater than the offset threshold is taken as the invalid spectral characteristic wavelength, and the invalid spectral characteristic wavelengths are eliminated to obtain all the valid spectral characteristic wavelengths; then, all the valid spectral characteristic wavelengths are taken as the characteristic wavelengths of the liquor and the base wine.

[0080] In some embodiments, a second prediction of the blending ratio between the liquor and the base liquor is performed based on the selected characteristic wavelengths to obtain the second blending parameter of the intelligent blending model, which can be achieved by the following steps:

[0081] Normalizing the selected characteristic wavelength to obtain a normalized characteristic wavelength;

[0082] Constructing a liquor blending prediction model according to the normalized characteristic wavelength;

[0083] The second blending parameter of the intelligent blending model is predicted by the liquor blending prediction model.

[0084] In specific implementation, first, the minimum-maximum normalization technology in the prior art can be used to normalize the screened characteristic wavelengths, and the data obtained after the normalization process is used as the normalized characteristic wavelength; then, a machine learning algorithm (for example: convolutional neural network) can be used to learn the correlation between the characteristic wavelength and the blending ratio, so as to pre-train the liquor blending model, and further input the normalized characteristic wavelength into the pre-trained liquor blending model for initialization, and then the characteristic wavelength is identified by the liquor blending model, and the blending ratio between the liquor and the base wine is mapped out, so as to complete the training of the liquor blending prediction model; finally, the blending ratio output by the liquor blending prediction model is used as the second blending parameter of the intelligent blending model; it should be noted that the second blending parameter in the present application refers to the blending ratio predicted based on the characteristic wavelength of the near-infrared spectrum of the liquor and the base wine.

[0085] In step 105, based on the first blending parameter and the second blending parameter, the intelligent blending model is coupled trained, and then the blending ratio between the liquor and the base liquor is predicted by the trained intelligent blending model.

[0086] In some embodiments, based on the first blending parameter and the second blending parameter, coupling training of the intelligent blending model can be implemented by the following steps:

[0087] Determining a coupling correlation between the first blending parameter and the second blending parameter;

[0088] Determining a sharing weight between two branch networks in the intelligent blending model according to the coupling correlation;

[0089] The first blending parameter and the second blending parameter are respectively used as prior parameters of two branch networks in the intelligent blending model, and then the intelligent blending model is trained using a preset training data set.

[0090] It should be noted that in this application, through the coupling training of the first blending parameter (based on the flavor component content) and the second blending parameter (based on the near-infrared spectrum), it is possible to integrate chemical and physical characteristics, quantify the coupling correlation between the two, and then optimize the shared weights of the dual-branch network in the intelligent blending model. This process utilizes the complementary advantages of different data sources. By taking the two parameters as prior knowledge, the learning ability and prediction accuracy of the model are further improved, ensuring the balance between flavor and molecular structure in the liquor blending process, and achieving more accurate blending ratio prediction.

[0091] It should also be noted that the coupling correlation in the present application is an indicator for measuring the degree of correlation between the first blending parameter and the second blending parameter. The smaller the coupling correlation, the greater the correlation between the first blending parameter and the second blending parameter, and the larger the coupling correlation, the smaller the correlation between the first blending parameter and the second blending parameter; in addition, the value range of the coupling correlation in the present application is between 0 and 1.

[0092] Preferably, in the above embodiment, determining the coupling correlation between the first blending parameter and the second blending parameter may be implemented by the following steps:

[0093] Obtaining the flavor component contents of liquor and base liquor, and then determining the content variance of the flavor component contents;

[0094] Acquire near infrared spectral data of liquor and base liquor, and then determine the spectral variance of the near infrared spectral data;

[0095] Determine the fluctuation correlation of the content determination of the compound component in the sample through the content variance and the spectral variance;

[0096] Determining a parameter difference between the first blending parameter and the second blending parameter;

[0097] The fluctuation correlation is compensated by the parameter difference, and the compensated fluctuation correlation is used as the coupling correlation between the first blending parameter and the second blending parameter.

[0098] In specific implementation, first, the flavor component contents of the liquor and the base wine are obtained by the method of the above-mentioned embodiment, and then the variance of the flavor component content corresponding to the liquor is calculated, and the variance is used as the liquor variance, and the variance of the flavor component content corresponding to the base wine is further calculated, and the variance is used as the base wine variance, and then the average value of the liquor variance and the base wine variance is used as the content variance of the flavor component content; secondly, the near-infrared spectral data of the liquor and the base wine are obtained by the method of the above-mentioned embodiment, the spectral variance of the liquor is calculated by the near-infrared spectral data corresponding to the liquor, and the spectral variance of the liquor is used as the spectral variance of the liquor, the spectral variance of the base wine is calculated by the near-infrared spectral data corresponding to the base wine, and the spectral variance of the base wine is used as the spectral variance of the base wine, and further, the average value of the spectral variance of the liquor and the spectral variance of the base wine can be used as the near-infrared spectral data The spectral variance of the data; secondly, the cosine similarity between the content variance and the spectral variance can be used as the fluctuation correlation of the compound component content determination in the sample; then, the absolute difference between the first blending parameter and the second blending parameter can be used as the parameter difference; finally, the natural exponential of the inverse of the parameter difference can be used as a compensation factor, and then the fluctuation correlation is compensated by the compensation factor, which is beneficial to reduce the systematic error introduced by the measurement, thereby improving the accuracy of the coupling correlation. As a preferred embodiment, compensating the fluctuation correlation by the compensation factor can be achieved in the following manner, namely: the compensation factor and the fluctuation correlation can be multiplied, and the product obtained by the multiplication is used as the coupling correlation between the first blending parameter and the second blending parameter.

[0099] In specific implementation, the shared weight between the two branch networks in the intelligent blending model is determined according to the coupling correlation, which can be implemented in the following manner, namely: the coupling correlation can be used to evaluate the correlation between features, that is, the magnitude of the coupling correlation can be used as the shared weight between the two branch networks in the intelligent blending model. For example, when the coupling correlation is 0.85, the shared weight of the intelligent blending model can be initialized to 0.85, and the shared weight vector W is [0.85, 0.85]. The intelligent blending model contains two branch networks (i.e., Branch A and Branch B), wherein each branch network accepts different input features, for example, the first branch network accepts the flavor component content data of the liquor and the base wine, and the second branch network accepts the near-infrared spectral data of the liquor and the base wine; the first blending parameter and the second blending parameter are respectively used as the prior parameters of the two branch networks in the intelligent blending model, and then the intelligent blending model is trained by a preset training data set. The following method can be adopted, namely: first, the first blending parameter is used as the prior parameter of the first branch network, and the second blending parameter is used as the prior parameter of the second branch network. It should be noted that the prior parameters in this application can be used as the initial constraints of the network to guide the model training direction; then, the flavor component content data in the preset training data set (i.e., the flavor component content of the liquor and the base wine) is used as the input of the first branch network in the intelligent blending model, and the near-infrared spectral data in the preset training data set (i.e., the near-infrared spectral data of the liquor and the base wine) is used as the input of the second branch network in the intelligent blending model. The two branch networks are then used to learn and optimize their respective blending characteristics, and are coupled through a shared layer. A suitable loss function (such as mean square error) is used to optimize the shared parameters and minimize the prediction error. When further evaluating the model performance, the validation set is used for testing to ensure the generalization ability of the model, thereby completing the training of the intelligent blending model based on the twin neural network.

[0100] In specific implementation, the blending ratio between the liquor and the base liquor is predicted by the trained intelligent blending model in the following manner, namely: in the intelligent blending model, the first branch network receives the flavor component content data corresponding to the first blending parameter, and the second branch network receives the near-infrared spectrum data corresponding to the second blending parameter. The intelligent blending model processes and analyzes the input features through its shared weights, calculates the output of each branch, and then shares and couples the outputs of the two branches to generate the blending ratio between the liquor and the base liquor.

[0101] It should be noted that the present application scheme constructs an intelligent blending model based on a twin neural network, comprehensively utilizes flavor components and near-infrared spectral data, and generates a first blending parameter and a second blending parameter respectively, thereby realizing a fusion analysis of multi-dimensional information. Through projection clustering and characteristic wavelength screening, the model not only optimizes the understanding of chemical and physical characteristics, but also improves the prediction accuracy in coupled training. Ultimately, the scheme effectively improves the accuracy of blending ratio prediction in the blending process of liquor, ensures the consistency of flavor and quality of the finished liquor, and meets the market demand for high-quality liquor.

[0102] On the other hand, in some embodiments, the present application provides a liquor intelligent blending system based on machine learning, the liquor intelligent blending system based on machine learning includes a liquor intelligent blending model training unit, reference Figure 4 , which is a schematic diagram of the structure of the liquor intelligent blending model training unit in some embodiments of the present application. The liquor intelligent blending model training unit 400 specifically includes: a construction module 401, a processing module 402 and an execution module 403, which are respectively described as follows:

[0103] Construction module 401. In this application, construction module 401 is mainly used to construct an intelligent blending model for liquor blending based on twin neural network initialization;

[0104] Processing module 402, in the present application, is used to collect the flavor component contents of the liquor and the base wine, perform an initial prediction of the blending ratio between the liquor and the base wine according to the flavor component contents of the liquor and the base wine, and obtain the first blending parameter of the intelligent blending model;

[0105] The processing module 402 in the present application is also used to collect near-infrared spectral data of the liquor and base liquor, perform projection clustering on the near-infrared spectral data to construct a spectral feature space, and then determine the baseline offset of each spectral feature wavelength according to the difference gradient between each spectral feature wavelength in the spectral feature space;

[0106] The processing module 402 in the present application is also used to filter out the characteristic wavelengths of the liquor and the base wine from the spectral feature space through the baseline offset of each spectral characteristic wavelength, and then perform a secondary prediction on the blending ratio between the liquor and the base wine based on the filtered characteristic wavelengths to obtain the second blending parameter of the intelligent blending model;

[0107] Execution module 403. In the present application, execution module 403 is mainly used to perform coupling training on the intelligent blending model based on the first blending parameter and the second blending parameter, and then predict the blending ratio between the liquor and the base liquor through the trained intelligent blending model.

[0108] In addition, the present application also provides a computer device, which includes a memory and a processor, the memory stores code, and the processor is configured to obtain the code and execute the above-mentioned machine learning-based intelligent liquor blending model training method.

[0109] In some embodiments, reference Figure 5 , which is a schematic diagram of the structure of a computer device for implementing a white wine intelligent blending model training method based on machine learning in some embodiments of the present application. The white wine intelligent blending model training method based on machine learning in the above embodiments can be Figure 5 The computer device 500 shown in the figure is implemented, and the computer device 500 includes at least one processor 501, a communication bus 502, a memory 503 and at least one communication interface 504.

[0110] The processor 501 may be a general-purpose central processing unit (CPU) or an application-specific integrated circuit (ASIC).

[0111] The communication bus 502 may be used to transmit information between the above-mentioned components.

[0112] The memory 503 may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a disk or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory 503 may exist independently and be connected to the processor 501 via the communication bus 502. The memory 503 may also be integrated with the processor 501.

[0113] Among them, the memory 503 is used to store the program code for executing the solution of the present application, and the execution is controlled by the processor 501. The processor 501 is used to execute the program code stored in the memory 503. The program code may include one or more software modules. The white wine intelligent blending model training method based on machine learning in the above embodiment can be implemented by the processor 501 and one or more software modules in the program code in the memory 503.

[0114] The communication interface 504 uses any transceiver or other device for communicating with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area networks (WLAN), etc.

[0115] In a specific implementation, as an embodiment, a computer device may include multiple processors, each of which may be a single-CPU processor or a multi-CPU processor. The processor here may refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0116] The above-mentioned computer device can be a general-purpose computer device or a special-purpose computer device. In a specific implementation, the computer device can be a desktop computer, a portable computer, a network server, a personal digital assistant (PDA), a mobile phone, a tablet computer, a wireless terminal device, a communication device or an embedded device. The embodiment of the present application does not limit the type of computer device.

[0117] In addition, the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the above-mentioned machine learning-based intelligent liquor blending model training method.

[0118] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0119] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A method for training a liquor intelligent blending model based on machine learning, which is used in a liquor intelligent blending system for liquor intelligent blending, characterized in that: The steps include: Construct an intelligent blending model for liquor blending based on twin neural network initialization; Collecting the flavor component contents of the liquor and the base wine, and making an initial prediction of the blending ratio between the liquor and the base wine according to the flavor component contents of the liquor and the base wine, to obtain a first blending parameter of the intelligent blending model; Collecting near-infrared spectral data of the liquor and base wine, performing projection clustering on the near-infrared spectral data to construct a spectral feature space, and then determining the baseline offset of each spectral feature wavelength according to the difference gradient between each spectral feature wavelength in the spectral feature space; The characteristic wavelengths of the liquor and the base wine are screened out from the spectral feature space by the baseline offset of each spectral characteristic wavelength, and then the blending ratio between the liquor and the base wine is secondary predicted based on the screened characteristic wavelengths to obtain the second blending parameter of the intelligent blending model; Based on the first blending parameter and the second blending parameter, the intelligent blending model is coupled trained, and then the blending ratio between the liquor and the base liquor is predicted by the trained intelligent blending model.

2. The method according to claim 1, characterized in that The blending ratio between the liquor and the base wine is initially predicted according to the flavor component content of the liquor and the base wine, and the first blending parameter of the intelligent blending model is obtained, which specifically includes: Standardizing the flavor component contents of the liquor and base liquor to obtain standard flavor component content data; The initial blending ratio of the liquor and the base liquor is predicted according to the standard flavor component content data, and the initial blending ratio is used as the first blending parameter of the intelligent blending model.

3. The method according to claim 1, characterized in that The projecting and clustering of the near infrared spectral data to construct a spectral feature space specifically includes: Extracting the main features of the near-infrared spectral data using principal component analysis to obtain a spectral feature set; Performing band hierarchical clustering on the spectral features in the spectral feature set to obtain spectral feature clusters of multiple bands; All spectral feature clusters are projected and mapped into a unified feature space to obtain a spectral feature space.

4. The method according to claim 1, characterized in that Determining the baseline offset of each spectral feature wavelength according to the difference gradient between each spectral feature wavelength in the spectral feature space specifically includes: Determine the difference gradient between the wavelengths of each spectral feature in the spectral feature space using a difference method; Based on a normal distribution assumption, determining a baseline wavelength of the spectral feature in the spectral feature space according to all difference gradients; The baseline offset of the corresponding spectral characteristic wavelength is determined by the baseline wavelength and each spectral characteristic wavelength.

5. The method according to claim 1, characterized in that Screening out the characteristic wavelengths of the liquor and base wine from the spectral feature space by using the baseline offset of each spectral characteristic wavelength specifically includes: Performing offset threshold comparison on the baseline offset of each spectral feature wavelength; When the baseline offset is less than or equal to a preset offset threshold, the spectral characteristic wavelength corresponding to the baseline offset is taken as the effective spectral characteristic wavelength, and then all effective spectral characteristic wavelengths are obtained; All effective spectral characteristic wavelengths are used as characteristic wavelengths of the liquor and base wine.

6. The method according to claim 1, characterized in that Based on the first blending parameter and the second blending parameter, coupling training of the intelligent blending model specifically includes: Determining a coupling correlation between the first blending parameter and the second blending parameter; Determining a sharing weight between two branch networks in the intelligent blending model according to the coupling correlation; The first blending parameter and the second blending parameter are respectively used as prior parameters of two branch networks in the intelligent blending model, and then the intelligent blending model is trained using a preset training data set.

7. The method according to claim 1, characterized in that The flavor component content refers to the specific concentration of various volatile compounds in the liquor and base wine.

8. A liquor intelligent blending system based on machine learning, comprising a liquor intelligent blending model training unit, characterized in that: The liquor intelligent blending model training unit comprises: A construction module is used to construct an intelligent blending model for liquor blending based on the twin neural network initialization; A processing module, for collecting the flavor component contents of the liquor and the base wine, and making an initial prediction of the blending ratio between the liquor and the base wine according to the flavor component contents of the liquor and the base wine, so as to obtain a first blending parameter of the intelligent blending model; The processing module is further used to collect near-infrared spectral data of the liquor and base wine, perform projection clustering on the near-infrared spectral data to construct a spectral feature space, and then determine the baseline offset of each spectral feature wavelength according to the difference gradient between each spectral feature wavelength in the spectral feature space; The processing module is further used to filter out the characteristic wavelengths of the liquor and the base wine from the spectral feature space by using the baseline offset of each spectral characteristic wavelength, and then perform a secondary prediction on the blending ratio between the liquor and the base wine based on the filtered characteristic wavelengths to obtain a second blending parameter of the intelligent blending model; An execution module is used to perform coupling training on the intelligent blending model based on the first blending parameter and the second blending parameter, and then predict the blending ratio between the liquor and the base liquor through the trained intelligent blending model.

9. A computer device, comprising a memory and a processor, wherein the memory stores a code, characterized in that: The processor is configured to obtain the code and execute the white wine intelligent blending model training method based on machine learning as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the machine learning-based liquor intelligent blending model training method as described in any one of claims 1 to 7 is implemented.

Images