Wine yield characteristic prediction method and intelligent batching method and device

Abstract

The invention relates to the technical field of wine making, and provides a wine yield characteristic prediction method and an intelligent batching method and device for ensuring the wine yield and quality while avoiding resource waste, the wine yield is predicted by establishing an in-cellar fermented grain physical and chemical characteristic prediction model and a wine yield characteristic prediction model, the brewing trial and error are effectively avoided, and the wine yield and quality are ensured. And resource waste is avoided and the efficiency is higher. According to the technical scheme, during batching, the target wine yield characteristics are set, the recommended in-cellar fermented grain batching characteristics can be obtained by adopting the optimization algorithm based on the wine yield characteristic prediction model, the parameter determination efficiency is higher, and the wine yield and quality after brewing better meet the requirements.

Description

Technical Field

[0001] The present invention relates to the technical field of brewing, and specifically to a method for predicting the characteristics of liquor production amount, an intelligent batching method and device. Background Art

[0002] The production process of Luzhou-flavor liquor is complex, involving multiple links such as the ecological environment, batching technology, and fermented grains fermentation. The traditional liquor production method mainly relies on the experience of the brewer and the intuitive judgment of the fermented grains. In the face of complex technological conditions and changing production environments, it is difficult to guarantee the liquor production amount and quality. Problems such as resource waste, fluctuations in output and quality may occur during the production process, affecting production efficiency and economic benefits. Summary of the Invention

[0003] In order to avoid resource waste while ensuring the liquor production amount and quality, the present application provides a method for predicting the characteristics of liquor production amount, an intelligent batching method and device.

[0004] The technical solution adopted by the present invention to solve the above problems is as follows:

[0005] A method for predicting the characteristics of liquor production amount, comprising:

[0006] Step 1, constructing a model training database, including ecological characteristics, physical and chemical characteristics of the fermented grains taken out of the cellar in the first round, batching characteristics of the fermented grains put into the cellar in the first round, physical and chemical characteristics of the fermented grains put into the cellar in the first round, and the corresponding liquor production amount characteristics of the second round;

[0007] Step 2, constructing a physical and chemical characteristics prediction model of the fermented grains put into the cellar based on the ecological characteristics, physical and chemical characteristics of the fermented grains taken out of the cellar in the first round, batching characteristics of the fermented grains put into the cellar in the first round, and physical and chemical characteristics of the fermented grains put into the cellar in the first round;

[0008] Step 3, constructing a liquor production amount characteristics prediction model based on the ecological characteristics, physical and chemical characteristics of the fermented grains taken out of the cellar in the first round, batching characteristics of the fermented grains put into the cellar in the first round, physical and chemical characteristics of the fermented grains put into the cellar in the first round, and the corresponding liquor production amount characteristics of the second round;

[0009] Step 4, determining the ecological characteristics, physical and chemical characteristics of the fermented grains taken out of the cellar in the current round, and batching characteristics of the fermented grains put into the cellar in the current round, inputting them into the physical and chemical characteristics prediction model of the fermented grains put into the cellar to obtain the predicted physical and chemical characteristics of the fermented grains put into the cellar in the current round, and inputting the ecological characteristics, physical and chemical characteristics of the fermented grains taken out of the cellar in the current round, batching characteristics of the fermented grains put into the cellar in the current round, and the predicted physical and chemical characteristics of the fermented grains put into the cellar in the current round into the liquor production amount characteristics prediction model to obtain the predicted liquor production amount characteristics of the next round.

[0010] Further, the model training database further includes the physical and chemical characteristics of the fermented grains taken out of the cellar in the second round;

[0011] The predicted model of liquor production volume characteristics includes a predicted unit of the physical and chemical characteristics of the fermented grains taken out of the cellar and a predicted unit of liquor production volume characteristics. The predicted unit of the physical and chemical characteristics of the fermented grains taken out of the cellar takes the ecological characteristics, the physical and chemical characteristics of the fermented grains taken out of the cellar in the first round, the ingredient characteristics of the fermented grains put into the cellar in the first round, and the physical and chemical characteristics of the fermented grains put into the cellar in the first round as inputs, and takes the physical and chemical characteristics of the fermented grains taken out of the cellar in the second round as outputs. The predicted unit of liquor production volume characteristics takes the ecological characteristics, the physical and chemical characteristics of the fermented grains taken out of the cellar in the first round, the ingredient characteristics of the fermented grains put into the cellar in the first round, the physical and chemical characteristics of the fermented grains put into the cellar in the first round, and the physical and chemical characteristics of the fermented grains taken out of the cellar in the second round as inputs, and takes the liquor production volume characteristics in the second round as outputs.

[0012] Furthermore, the ecological characteristics include: the ground temperature, the temperature of the fermented grains put into the cellar, the air temperature, and the air humidity;

[0013] The ingredient characteristics of the fermented grains put into the cellar include: the water amount for moistening grains, the grain amount, the bran amount, the water amount for sizing, the koji amount, the steaming duration of grains, and the layer of the fermented grains;

[0014] The physical and chemical characteristics of the fermented grains put into the cellar include: the moisture content, the acidity, and the starch content of the fermented grains put into the cellar;

[0015] The physical and chemical characteristics of the fermented grains taken out of the cellar include: the moisture content, the acidity, the starch content, and the residual sugar content of the fermented grains taken out of the cellar;

[0016] The liquor production volume characteristics include: the grade and the production volume of the liquor produced.

[0017] Furthermore, it also includes step 5, modifying the predicted model of liquor production volume characteristics based on the difference between the predicted liquor production volume characteristics obtained in step 3 and the actual liquor production volume characteristics.

[0018] The device for predicting liquor production volume characteristics includes:

[0019] The database construction module is used to construct a model training database, including ecological characteristics, the physical and chemical characteristics of the fermented grains taken out of the cellar in the first round, the ingredient characteristics of the fermented grains put into the cellar in the first round, the physical and chemical characteristics of the fermented grains put into the cellar in the first round, and the corresponding liquor production volume characteristics in the second round;

[0020] The model construction module constructs and trains the predicted model of the physical and chemical characteristics of the fermented grains put into the cellar and the predicted model of liquor production volume characteristics based on the model training database;

[0021] The prediction module is used to determine the ecological characteristics, the physical and chemical characteristics of the fermented grains taken out of the cellar in the current round, the ingredient characteristics of the fermented grains put into the cellar in the current round, the physical and chemical characteristics of the fermented grains put into the cellar in the current round, and obtain the predicted liquor production volume characteristics in the next round based on the predicted model of the physical and chemical characteristics of the fermented grains put into the cellar and the predicted model of liquor production volume characteristics.

[0022] The intelligent batching method includes:

[0023] Step 1: Construct a model training database, including ecological characteristics, physical and chemical characteristics of the fermented grains taken out of the cellar in the first round, ingredient characteristics of the fermented grains put into the cellar in the first round, physical and chemical characteristics of the fermented grains put into the cellar in the first round, and the corresponding liquor yield characteristics in the second round;

[0024] Step 2: Based on the ecological characteristics, physical and chemical characteristics of the fermented grains taken out of the cellar in the first round, ingredient characteristics of the fermented grains put into the cellar in the first round, and physical and chemical characteristics of the fermented grains put into the cellar in the first round, construct a physical and chemical characteristics prediction model for the fermented grains put into the cellar;

[0025] Step 3: Based on the ecological characteristics, physical and chemical characteristics of the fermented grains taken out of the cellar in the first round, ingredient characteristics of the fermented grains put into the cellar in the first round, physical and chemical characteristics of the fermented grains put into the cellar in the first round, and the corresponding liquor yield characteristics in the second round, construct a liquor yield characteristics prediction model;

[0026] Step 4: Based on the physical and chemical characteristics prediction model for the fermented grains put into the cellar and the liquor yield characteristics prediction model, construct an intelligent batching model;

[0027] Step 5: Determine the ecological characteristics, physical and chemical characteristics of the fermented grains taken out of the cellar in the current round, and the target liquor yield characteristics in the next round; Initialize the ingredient characteristics of the fermented grains put into the cellar in the current round, and input the ecological characteristics, physical and chemical characteristics of the fermented grains taken out of the cellar in the current round, ingredient characteristics of the fermented grains put into the cellar in the current round, and the target liquor yield characteristics in the next round into the intelligent batching model for optimization to obtain the recommended ingredient characteristics of the fermented grains put into the cellar in the current round.

[0028] Furthermore, the model training database also includes the physical and chemical characteristics of the fermented grains taken out of the cellar in the second round;

[0029] The liquor yield characteristics prediction model includes a physical and chemical characteristics prediction unit for the fermented grains taken out of the cellar and a liquor yield characteristics prediction unit. The physical and chemical characteristics prediction unit for the fermented grains taken out of the cellar takes the ecological characteristics, physical and chemical characteristics of the fermented grains taken out of the cellar in the first round, ingredient characteristics of the fermented grains put into the cellar in the first round, and physical and chemical characteristics of the fermented grains put into the cellar in the first round as inputs, and outputs the physical and chemical characteristics of the fermented grains taken out of the cellar in the second round. The liquor yield characteristics prediction unit takes the ecological characteristics, physical and chemical characteristics of the fermented grains taken out of the cellar in the first round, ingredient characteristics of the fermented grains put into the cellar in the first round, physical and chemical characteristics of the fermented grains put into the cellar in the first round, and the physical and chemical characteristics of the fermented grains taken out of the cellar in the second round as inputs, and outputs the liquor yield characteristics in the second round.

[0030] Furthermore, Step 5 also includes determining the target physical and chemical characteristics of the fermented grains taken out of the cellar in the next round.

[0031] Furthermore, Step 5 performs optimization based on the genetic algorithm, simulated annealing algorithm, or evolutionary strategy algorithm.

[0032] Intelligent batching device, including:

[0033] A database construction module for constructing a model training database, including ecological characteristics, physical and chemical characteristics of the fermented grains taken out of the cellar in the first round, ingredient characteristics of the fermented grains put into the cellar in the first round, physical and chemical characteristics of the fermented grains put into the cellar in the first round, and the corresponding liquor yield characteristics in the second round.

[0034] A model construction module for constructing a physical and chemical characteristics prediction model of the fermented grains put into the cellar and a liquor yield characteristics prediction model based on the model training database, and constructing an intelligent ingredient model based on the physical and chemical characteristics prediction model of the fermented grains put into the cellar and the liquor yield characteristics prediction model.

[0035] A recommendation module for obtaining the ingredient characteristics of the fermented grains put into the cellar in the current round to be recommended by using the intelligent ingredient model according to the determined ecological characteristics, physical and chemical characteristics of the fermented grains taken out of the cellar in the current round, and the target liquor yield characteristics in the next round.

[0036] The beneficial effects of the present invention compared with the prior art are:

[0037] The liquor yield characteristics prediction method provided by the present application includes two parts: physical and chemical characteristics prediction of the fermented grains put into the cellar and liquor yield characteristics prediction. The physical and chemical characteristics prediction of the fermented grains put into the cellar can obtain the physical and chemical characteristics of the fermented grains put into the cellar in the current round before the physical and chemical characteristics of the fermented grains taken out of the cellar in the current round are mixed with the ingredient characteristics of the fermented grains put into the cellar in the current round, without the need to obtain the physical and chemical characteristics of the fermented grains put into the cellar by actual measurement, which can effectively avoid waste of resources; through the liquor yield characteristics prediction, it can be known in advance whether the current ingredients meet the liquor yield requirements, and only when the liquor yield requirements are met, the formal brewing is carried out, effectively avoiding brewing trial and error, avoiding waste of resources and having higher efficiency. In order to further avoid waste of resources, the present application also conducts physical and chemical characteristics prediction of the fermented grains taken out of the cellar, ensuring the liquor yield and quality, and ensuring that the physical and chemical characteristics of the fermented grains taken out of the cellar meet the requirements for brewing use.

[0038] The present application also provides an intelligent ingredient method. By setting the target liquor yield characteristics, the recommended ingredient characteristics of the fermented grains put into the cellar can be obtained by using an optimization algorithm based on the liquor yield characteristics prediction model, with higher parameter determination efficiency, and the liquor yield and quality after brewing more meet the requirements. Description of the Drawings

[0039] Figure 1 It is a flowchart of the liquor yield characteristics prediction method;

[0040] Figure 2 It is a schematic diagram of the prediction result of the acidity parameter in the physical and chemical parameters of the fermented grains taken out of the cellar by the physical and chemical characteristics prediction unit of the fermented grains taken out of the cellar;

[0041] Figure 3 It is a schematic diagram of the structure of the liquor yield characteristics prediction device;

[0042] Figure 4 It is a flowchart of the intelligent ingredient method;

[0043] Figure 5 It is a schematic structural diagram of an intelligent batching device. Specific implementation manners

[0044] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in conjunction with embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0045] The rounds in the liquor production process can be roughly divided into: Round 1: taking out the fermented grains - batching - liquor production - putting into the cellar for fermentation; Round 2: taking out the fermented grains - batching - liquor production - putting into the cellar for fermentation;...

[0046] The liquor production effect stems from the fermentation effect of the fermented grains, which is determined by the fermentation of the fermented grains in the previous round. Therefore, when predicting the liquor production amount characteristics in this embodiment, the taking out of the cellar, batching and putting into the cellar in the current round are used to predict the liquor production amount characteristics in the next round; in order to improve the prediction accuracy, in addition to the taking out of the cellar, batching and putting into the cellar in the current round, the taking out of the cellar in the next round is also added to predict the liquor production amount characteristics in the next round.

[0047] In order to facilitate the distinction between the model establishment process and the model use process, during the model establishment, "the first round" and "the second round" are used for description, and during the model use, "the current round" and "the next round" are used for description.

[0048] As Figure 1 shown, the method for predicting the liquor production amount characteristics includes:

[0049] Step 1, construct a model training database, including ecological characteristics, physical and chemical characteristics of the fermented grains taken out of the cellar in the first round, batching characteristics of the fermented grains put into the cellar in the first round, physical and chemical characteristics of the fermented grains put into the cellar in the first round, and the corresponding liquor production amount characteristics in the second round.

[0050] The ecological characteristics include: ground temperature, temperature of the fermented grains put into the cellar, air temperature and air humidity;

[0051] The batching characteristics of the fermented grains put into the cellar include: water amount for moistening grains added to the fermented grains, grain amount added, bran amount added, water amount for sizing, koji amount added, steaming time of grains, and layers of the fermented grains (divided into upper, middle and lower layers);

[0052] The physical and chemical characteristics of the fermented grains put into the cellar include: moisture content, acidity and starch content of the fermented grains put into the cellar;

[0053] The physical and chemical characteristics of the fermented grains taken out of the cellar include: moisture content, acidity, starch content and residual sugar content of the fermented grains taken out of the cellar;

[0054] The liquor production amount characteristics include: grade and output of the liquor produced.

[0055] Ecological characteristics, characteristics of the fermented grains charged into the cellar, physical and chemical characteristics of the fermented grains charged into the cellar, physical and chemical characteristics of the fermented grains taken out of the cellar, and characteristics of liquor yield may also include other characteristics concerned by experts or engineers in the field.

[0056] In this embodiment, the fermented grains layer data is encoded using one-hot vectors. For example, the upper layer is represented as [1, 0, 0], the middle layer is represented as [0, 1, 0], and the lower layer is represented as [0, 0, 1]; the remaining data uses conventional units, and when collecting each data, it is only necessary to keep the units of each index consistent.

[0057] To improve the data processing efficiency, the data in the model training database can also be normalized, such as using min-max normalization or Z-score normalization, etc.

[0058] The data in the model training database can be updated and supplemented according to actual needs to ensure the accuracy of the established model.

[0059] Step 2: Construct a physical and chemical characteristics prediction model of the fermented grains charged into the cellar based on ecological characteristics, physical and chemical characteristics of the fermented grains taken out of the cellar in the first round, ingredients characteristics of the fermented grains charged into the cellar in the first round, and physical and chemical characteristics of the fermented grains charged into the cellar in the first round.

[0060] The physical and chemical characteristics prediction model of the fermented grains charged into the cellar is constructed based on neural networks such as RNN, LSTM, and MLP that can be used to process one-dimensional data; using ecological characteristics, physical and chemical characteristics of the fermented grains taken out of the cellar in the first round, and ingredients characteristics of the fermented grains charged into the cellar in the first round as input features, and using the physical and chemical characteristics of the fermented grains charged into the cellar in the first round as the output label, construct a sample data set, divide the sample data set into a training set and a test set, train the neural network model with the training set, during the training process, compare the predicted value of the neural network model with the corresponding true output label, calculate the corresponding loss function (such as SmoothL1Loss), and optimize the model through the backpropagation of the neural network; verify the neural network model with the test set, when the corresponding loss function is less than the loss function threshold, use the corresponding neural network model as the physical and chemical characteristics prediction model of the fermented grains charged into the cellar.

[0061] By establishing a physical and chemical characteristics prediction model of the fermented grains charged into the cellar, it is possible to predict the physical and chemical characteristics of the fermented grains charged into the cellar without using the actual measurement method to obtain the physical and chemical characteristics of the fermented grains charged into the cellar, and save resources by avoiding the ineffective mixing of the physical and chemical characteristics of the fermented grains taken out of the cellar and the ingredients characteristics of the fermented grains charged into the cellar.

[0062] Step 3: Construct a liquor yield characteristics prediction model based on ecological characteristics, physical and chemical characteristics of the fermented grains taken out of the cellar in the first round, ingredients characteristics of the fermented grains charged into the cellar in the first round, physical and chemical characteristics of the fermented grains charged into the cellar in the first round, and the corresponding liquor yield characteristics in the second round.

[0063] The liquor production volume characteristic prediction model is also constructed based on neural networks such as RNN, LSTM, and MLP that can be used to process one-dimensional data. Using ecological characteristics, the physical and chemical characteristics of the fermented grains taken out of the cellar in the first round, the ingredient characteristics of the fermented grains put into the cellar in the first round, and the physical and chemical characteristics of the fermented grains put into the cellar in the first round as inputs, and the liquor production volume characteristics of the second round as outputs. The model training process is similar to that of the physical and chemical characteristics prediction model of the fermented grains put into the cellar, and will not be elaborated here.

[0064] By constructing the liquor production volume characteristic prediction model, it is possible to know in advance whether the current ingredients meet the liquor production volume requirements. Only when the liquor production volume requirements are met can the formal brewing be carried out, effectively avoiding brewing trial and error, avoiding waste of resources and being more efficient.

[0065] In order to further avoid waste of resources, in this embodiment, when constructing the model training database, the physical and chemical characteristics of the fermented grains taken out of the cellar in the second round are also added. The liquor production volume characteristic prediction model includes a physical and chemical characteristics prediction unit of the fermented grains taken out of the cellar and a liquor production volume characteristic prediction unit. The physical and chemical characteristics prediction unit of the fermented grains taken out of the cellar uses ecological characteristics, the physical and chemical characteristics of the fermented grains taken out of the cellar in the first round, the ingredient characteristics of the fermented grains put into the cellar in the first round, and the physical and chemical characteristics of the fermented grains put into the cellar in the first round as inputs, and the physical and chemical characteristics of the fermented grains taken out of the cellar in the second round as outputs. The liquor production volume characteristic prediction unit uses ecological characteristics, the physical and chemical characteristics of the fermented grains taken out of the cellar in the first round, the ingredient characteristics of the fermented grains put into the cellar in the first round, the physical and chemical characteristics of the fermented grains put into the cellar in the first round, and the physical and chemical characteristics of the fermented grains taken out of the cellar in the second round as inputs, and the liquor production volume characteristics of the second round as outputs.

[0066] As Figure 2 shown, it shows the comparison between the predicted value and the true value of the acidity parameter in the physical and chemical parameters of the fermented grains taken out of the cellar by the physical and chemical characteristics prediction unit of the fermented grains taken out of the cellar. It can be seen that R 2 is greater than 0.9, indicating that the prediction effect is good.

[0067] By predicting the physical and chemical characteristics of the fermented grains taken out of the cellar, while ensuring the liquor production volume and quality, it ensures that the physical and chemical characteristics of the fermented grains taken out of the cellar meet the requirements for brewing use, avoiding waste of resources.

[0068] Step 4: Determine the ecological characteristics, the physical and chemical characteristics of the fermented grains taken out of the cellar in the current round, and the ingredient characteristics of the fermented grains put into the cellar in the current round, input them into the physical and chemical characteristics prediction model of the fermented grains put into the cellar to obtain the predicted physical and chemical characteristics of the fermented grains put into the cellar in the current round, and input the ecological characteristics, the physical and chemical characteristics of the fermented grains taken out of the cellar in the current round, the ingredient characteristics of the fermented grains put into the cellar in the current round, and the predicted physical and chemical characteristics of the fermented grains put into the cellar in the current round into the liquor production volume characteristic prediction model to obtain the predicted liquor production volume characteristics of the next round.

[0069] When the predicted production volume feature model includes the predicted physical and chemical characteristics unit of the cellar-exited fermented grains and the predicted production volume feature unit, the predicted physical and chemical characteristics unit of the cellar-exited fermented grains first predicts the physical and chemical characteristics of the next-round cellar-exited fermented grains according to the input ecological characteristics, the physical and chemical characteristics of the current-round cellar-exited fermented grains, the ingredient characteristics of the current-round cellar-entered fermented grains, and the predicted physical and chemical characteristics of the current-round cellar-entered fermented grains; then, the predicted production volume feature unit predicts the production volume characteristics of the next round according to the ecological characteristics, the physical and chemical characteristics of the current-round cellar-exited fermented grains, the ingredient characteristics of the current-round cellar-entered fermented grains, the predicted physical and chemical characteristics of the current-round cellar-entered fermented grains, and the predicted physical and chemical characteristics of the next-round cellar-exited fermented grains.

[0070] To improve the accuracy of the prediction model, the prediction model can also be corrected according to the difference between the measured value and the predicted value. The corrected model includes the predicted physical and chemical characteristics model of the cellar-entered fermented grains and the predicted production volume feature model.

[0071] As Figure 3 shown, this embodiment correspondingly provides a predicted production volume feature device, including:

[0072] A database construction module, used to construct a model training database, including ecological characteristics, the physical and chemical characteristics of the cellar-exited fermented grains in the first round, the ingredient characteristics of the cellar-entered fermented grains in the first round, the physical and chemical characteristics of the cellar-entered fermented grains in the first round, and the corresponding production volume characteristics of the second round;

[0073] A model construction module, which constructs and trains the predicted physical and chemical characteristics model of the cellar-entered fermented grains and the predicted production volume feature model based on the model training database;

[0074] A prediction module, used to determine the ecological characteristics, the physical and chemical characteristics of the current-round cellar-exited fermented grains, the ingredient characteristics of the current-round cellar-entered fermented grains, the physical and chemical characteristics of the current-round cellar-entered fermented grains, and obtain the predicted production volume characteristics of the next round based on the predicted physical and chemical characteristics model of the cellar-entered fermented grains and the predicted production volume feature model.

[0075] As Figure 4 shown, this embodiment also provides an intelligent batching method, including:

[0076] Step 1, construct a model training database, including ecological characteristics, the physical and chemical characteristics of the cellar-exited fermented grains in the first round, the ingredient characteristics of the cellar-entered fermented grains in the first round, the physical and chemical characteristics of the cellar-entered fermented grains in the first round, and the corresponding production volume characteristics of the second round;

[0077] Step 2, construct a predicted physical and chemical characteristics model of the cellar-entered fermented grains based on the ecological characteristics, the physical and chemical characteristics of the cellar-exited fermented grains in the first round, the ingredient characteristics of the cellar-entered fermented grains in the first round, and the physical and chemical characteristics of the cellar-entered fermented grains in the first round;

[0078] Step 3: Construct a predicted production volume feature model based on ecological features, the physical and chemical features of the fermented grains taken out of the cellar in the first round, the ingredient features of the fermented grains put into the cellar in the first round, the physical and chemical features of the fermented grains put into the cellar in the first round, and the corresponding production volume features of the second round.

[0079] Steps 1 - 3 in the intelligent ingredient method are similar to Steps 1 - 3 in the predicted production volume feature method, and thus will not be elaborated here.

[0080] Step 4: Construct an intelligent ingredient model based on the predicted physical and chemical feature model of the fermented grains put into the cellar and the predicted production volume feature model.

[0081] Step 5: Determine the ecological features, the physical and chemical features of the fermented grains taken out of the cellar in the current round, and the target production volume features of the next round; initialize the ingredient features of the fermented grains put into the cellar in the current round, and input the ecological features, the physical and chemical features of the fermented grains taken out of the cellar in the current round, the ingredient features of the fermented grains put into the cellar in the current round, and the target production volume features of the next round into the intelligent ingredient model for optimization to obtain the recommended ingredient features of the fermented grains put into the cellar in the current round. The methods used for optimization include genetic algorithm, simulated annealing algorithm, or evolutionary strategy algorithm.

[0082] Take the genetic algorithm as an example to illustrate the optimization process:

[0083] Input the ecological features, the physical and chemical features of the fermented grains taken out of the cellar in the current round, and the initialized ingredient features of the fermented grains put into the cellar into the predicted physical and chemical feature model of the fermented grains put into the cellar to obtain the predicted physical and chemical features of the fermented grains put into the cellar.

[0084] Input the ecological features, the physical and chemical features of the fermented grains taken out of the cellar in the current round, the initialized ingredient features of the fermented grains put into the cellar, and the predicted physical and chemical features of the fermented grains put into the cellar into the predicted production volume feature model to obtain the predicted production volume features.

[0085] Calculate the fitness function based on the difference between the predicted production volume features and the target production volume features, and then adjust the ingredient features of the fermented grains put into the cellar through operations such as selection, crossover, and mutation of the genetic algorithm according to the fitness function value.

[0086] Then repeat the above steps until the difference between the predicted production volume features and the target production volume features meets the requirements. At this time, the ingredient features of the fermented grains put into the cellar are the recommended ingredient features of the fermented grains put into the cellar.

[0087] By setting the target production volume features, the recommended ingredient features of the fermented grains put into the cellar can be obtained based on the predicted production volume feature model using the optimization algorithm, with higher parameter determination efficiency, and the production volume and quality after brewing more meeting the requirements.

[0088] If the predicted model of liquor production volume characteristics takes into account the physical and chemical characteristics of the fermented grains taken out of the cellar, the physical and chemical characteristics of the target fermented grains taken out of the cellar can also be added during optimization. The specific optimization process is similar to the above method and will not be elaborated here.

[0089] As Figure 5 shown, this embodiment also provides an intelligent batching device corresponding to the intelligent batching method, including:

[0090] A database construction module for constructing a model training database, including ecological characteristics, physical and chemical characteristics of the fermented grains taken out of the cellar in the first round, batching characteristics of the fermented grains put into the cellar in the first round, physical and chemical characteristics of the fermented grains put into the cellar in the first round, and the corresponding liquor production volume characteristics in the second round;

[0091] A model construction module for constructing a predicted model of the physical and chemical characteristics of the fermented grains put into the cellar and a predicted model of the liquor production volume characteristics based on the model training database, and constructing an intelligent batching model based on the predicted model of the physical and chemical characteristics of the fermented grains put into the cellar and the predicted model of the liquor production volume characteristics;

[0092] A recommendation module for obtaining the batching characteristics of the fermented grains put into the cellar in the current round to be recommended by using the intelligent batching model according to the determined ecological characteristics, physical and chemical characteristics of the fermented grains taken out of the cellar in the current round, and the target liquor production volume characteristics in the next round.

Claims

1. A method for predicting wine production characteristics, characterized in that: include: Step 1, construct a model training database, including ecological characteristics, physical and chemical characteristics of the first round of uncellared mash, ingredient characteristics of the first round of cellared mash, physical and chemical characteristics of the first round of cellared mash, and corresponding second round wine production characteristics; Step 2, constructing a prediction model for the physicochemical characteristics of the fermented grains entering the cellar based on the ecological characteristics, the physicochemical characteristics of the first round of fermented grains out of the cellar, the ingredient characteristics of the first round of fermented grains entering the cellar, and the physicochemical characteristics of the first round of fermented grains entering the cellar; Step 3, constructing a wine yield characteristic prediction model based on ecological characteristics, physical and chemical characteristics of the first round of uncellared lees, ingredient characteristics of the first round of cellared lees, physical and chemical characteristics of the first round of cellared lees, and corresponding second round wine yield characteristics; Step 4, determine the ecological characteristics, the physical and chemical characteristics of the current round of uncellared mash, and the characteristics of the ingredients of the current round of mash entering the cellar, and input them into the physical and chemical characteristics prediction model of the mash entering the cellar to obtain the predicted physical and chemical characteristics of the current round of mash entering the cellar, and input the ecological characteristics, the physical and chemical characteristics of the current round of uncellared mash, the characteristics of the ingredients of the current round of mash entering the cellar, and the predicted physical and chemical characteristics of the current round of mash entering the cellar into the wine production characteristic prediction model to obtain the predicted wine production characteristics of the next round.

2. The method for predicting wine production characteristics according to claim 1, characterized in that: The model training database also includes the physical and chemical characteristics of the second round of uncellared mash; The wine yield characteristic prediction model includes a wine yield characteristic prediction unit and a wine yield characteristic prediction unit. The wine yield characteristic prediction unit takes the ecological characteristics, the physicochemical characteristics of the wine yield from the first round, the characteristics of the ingredients of the wine yield from the first round, and the physicochemical characteristics of the wine yield from the first round as input, and takes the physicochemical characteristics of the wine yield from the second round as output. The wine yield characteristic prediction unit takes the ecological characteristics, the physicochemical characteristics of the wine yield from the first round, the characteristics of the ingredients of the wine yield from the first round, the physicochemical characteristics of the wine yield from the first round, and the physicochemical characteristics of the wine yield from the second round as input, and takes the wine yield characteristics of the second round as output.

3. The method for predicting wine production characteristics according to claim 2, characterized in that: Ecological characteristics include: ground temperature, cellar temperature, air temperature, and air humidity; The characteristics of the ingredients of the fermented grains entering the cellar include: the amount of water added to moisten the fermented grains, the amount of grain added, the amount of bran added, the amount of water added, the amount of koji added, the steaming time and the layers of the fermented grains; The physical and chemical characteristics of the fermented grains entering the cellar include: moisture, acidity and starch content of the fermented grains entering the cellar; The physical and chemical characteristics of the uncellared mash include: moisture, acidity, starch content and residual sugar content; Wine production characteristics include: wine grade and output.

4. The method for predicting wine production characteristics according to any one of claims 1 to 3, characterized in that: The method further includes step 5, correcting the wine production characteristic prediction model based on the difference between the predicted wine production characteristic obtained in step 3 and the actual wine production characteristic.

5. A device for predicting wine production characteristics, characterized in that: include: A database construction module is used to construct a model training database, including ecological characteristics, physical and chemical characteristics of the first round of uncellared mash, ingredient characteristics of the first round of cellared mash, physical and chemical characteristics of the first round of cellared mash, and corresponding second round wine production characteristics; Model building module, based on the model training database, builds and trains the prediction model of the physical and chemical characteristics of the fermented grains entering the cellar and the prediction model of the wine production characteristics; The prediction module is used to determine the ecological characteristics, the physical and chemical characteristics of the current round of uncellared mash, the ingredient characteristics of the current round of mash entering the cellar, and the physical and chemical characteristics of the current round of mash entering the cellar, and obtain the predicted wine production characteristics of the next round based on the physical and chemical characteristics prediction model of the mash entering the cellar and the wine production characteristics prediction model.

6. Intelligent batching method, characterized in that: include: Step 1, construct a model training database, including ecological characteristics, physical and chemical characteristics of the first round of uncellared mash, ingredient characteristics of the first round of cellared mash, physical and chemical characteristics of the first round of cellared mash, and corresponding second round wine production characteristics; Step 2, constructing a prediction model for the physicochemical characteristics of the fermented grains entering the cellar based on the ecological characteristics, the physicochemical characteristics of the first round of fermented grains out of the cellar, the ingredient characteristics of the first round of fermented grains entering the cellar, and the physicochemical characteristics of the first round of fermented grains entering the cellar; Step 3, constructing a wine yield characteristic prediction model based on ecological characteristics, physical and chemical characteristics of the first round of uncellared lees, ingredient characteristics of the first round of cellared lees, physical and chemical characteristics of the first round of cellared lees, and corresponding second round wine yield characteristics; Step 4: construct an intelligent batching model based on the prediction model of the physicochemical characteristics of the fermented grains entering the cellar and the prediction model of the wine production characteristics; Step 5, determine the ecological characteristics, the physical and chemical characteristics of the current round of uncellared mash, and the target wine production characteristics of the next round; initialize the ingredients characteristics of the current round of mash entering the cellar, and input the ecological characteristics, the physical and chemical characteristics of the current round of uncellared mash, the ingredients characteristics of the current round of mash entering the cellar, and the target wine production characteristics of the next round into the intelligent ingredient model for optimization to obtain the recommended ingredients characteristics of the current round of mash entering the cellar.

7. The intelligent batching method according to claim 6, characterized in that: The model training database also includes the physical and chemical characteristics of the second round of uncellared mash; The wine yield characteristic prediction model includes a wine yield characteristic prediction unit and a wine yield characteristic prediction unit. The wine yield characteristic prediction unit takes the ecological characteristics, the physicochemical characteristics of the wine yield from the first round, the characteristics of the ingredients of the wine yield from the first round, and the physicochemical characteristics of the wine yield from the first round as input, and takes the physicochemical characteristics of the wine yield from the second round as output. The wine yield characteristic prediction unit takes the ecological characteristics, the physicochemical characteristics of the wine yield from the first round, the characteristics of the ingredients of the wine yield from the first round, the physicochemical characteristics of the wine yield from the first round, and the physicochemical characteristics of the wine yield from the second round as input, and takes the wine yield characteristics of the second round as output.

8. The intelligent batching method according to claim 7, characterized in that: Step 5 also includes determining the target physical and chemical characteristics of the next round of mash out of the cellar.

9. The intelligent batching method according to any one of claims 6 to 8, characterized in that: Step 5 is to perform optimization based on genetic algorithm, simulated annealing algorithm or evolution strategy algorithm.

10. Intelligent batching device, characterized in that: include: A database construction module is used to construct a model training database, including ecological characteristics, physical and chemical characteristics of the first round of uncellared mash, ingredient characteristics of the first round of cellared mash, physical and chemical characteristics of the first round of cellared mash, and corresponding second round wine production characteristics; A model building module, which builds a prediction model for the physicochemical characteristics of the fermented grains entering the cellar and a prediction model for the characteristics of the wine production based on the model training database, and builds an intelligent batching model based on the prediction model for the physicochemical characteristics of the fermented grains entering the cellar and the prediction model for the characteristics of the wine production; The recommendation module uses an intelligent ingredient model to obtain the recommended ingredient characteristics of the current round of mash to be put into the cellar, based on the determined ecological characteristics, the physical and chemical characteristics of the current round of mash out of the cellar, and the target wine production characteristics of the next round.

Images