Enzymatic performance evaluation and key microorganism identification method and device of yeast for making hard liquor
The contribution of microorganisms in Daqu was evaluated through neural network model and SHAP theory, and the problem of unclear influence of microorganisms in Daqu was solved, systematic evaluation of enzymatic performance and identification of key microorganisms were achieved, and the control ability of the fermentation process was improved.
Patent Information
- Application Number
- CN202510365714.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-26
- Publication Date
- 2025-07-08
AI Technical Summary
The impact of each microorganism in Daqu on the enzymatic performance is unclear, which makes it difficult to identify and regulate key microorganisms and lacks systematic and intelligent evaluation methods.
The neural network model is used to train the enzymatic performance prediction model, and the microbial characteristic contribution module is constructed in combination with SHAP theory. The contribution of each microbial characteristic is evaluated through the SHAP value of the microbial characteristics, key microorganisms are selected, and quantitative mapping relationship between the Daqu microbial community and the enzyme performance is established.
The systematic evaluation of Daqu zymological performance and the identification of key microorganisms are achieved, providing theoretical support and practical tools for quality control and fermentation optimization.
Smart Images
Figure CN120280047A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of brewing, and particularly relates to a method and device for evaluating the enzymatic properties of Daqu and identifying key microorganisms. Background Art
[0002] As an important raw material in the fermentation of traditional Chinese liquor, the enzymatic properties of Daqu directly affect the efficiency of the fermentation process and the quality of the product. At present, the evaluation of the enzymatic properties of Daqu mostly relies on experience and experimental determination, lacking systematic and intelligent evaluation methods. In addition, there are many types of microorganisms in Daqu, but the specific effects of each microorganism on the enzymatic properties are not clear, resulting in great difficulties in identifying and regulating key microorganisms during the production process. Summary of the Invention
[0003] The technical problem to be solved by the present invention: The present invention provides a method and device for evaluating the enzymatic properties of Daqu and identifying key microorganisms, which solves the problem that the effects of various microorganisms in existing Daqu on the enzymatic properties are not clear.
[0004] The technical solution adopted by the present invention to solve the above technical problem: A method for evaluating the enzymatic properties of Daqu and identifying key microorganisms includes the following steps:
[0005] S1. Obtain the microbial characteristics and enzymatic property characteristics of the Daqu sample;
[0006] S2. Establish a neural network model, use the microbial characteristics as the input and the enzymatic property characteristics as the output, train the neural network model to obtain an enzymatic property prediction model;
[0007] S3. Construct a microbial feature contribution module based on the SHAP theory, which is used to calculate the SHAP values of the microbial characteristics for the prediction results, and evaluate the contribution degree of each microbial characteristic according to the calculated SHAP values, and select microorganisms as key microorganisms based on the contribution degree;
[0008] S4. Obtain the microbial characteristics of the Daqu to be predicted, use the enzymatic property prediction model to obtain the prediction result of the enzymatic property characteristics of the Daqu to be measured, and use the microbial feature contribution module to obtain the key microorganism identification result.
[0009] Further, the microbial characteristics include the types and quantities of microbial populations during the Daqu fermentation process, and the types of the microbial populations include Bacillus, Aspergillus, Saccharomyces cerevisiae, Pediococcus acidilactici, and Staphylococcus.
[0010] Further, the enzymatic property characteristics include saccharifying power, fermenting power, liquefying power, and esterifying power.
[0011] Further, the neural network model includes one or a combination of more than one of MLP, LSTM, RNN, CNN, GRU, and Transformer.
[0012] Further, the calculation formula of the SHAP value is: where, φ j is the SHAP value of the j-th microorganism; N is the set of all microorganism features; N\{j} is all subsets excluding the j-th microorganism feature, S is a subset in N\{j}; v(S) is the output value of the enzymatic property prediction model when only subset S is included, and v(S∪{j}) - v(S) represents the change in the output value of the enzymatic property prediction model after adding the j-th microorganism feature; |S| is the size of subset S; |N| is the total number of microorganism features.
[0013] Further, the loss function of the enzymatic property prediction model is: where, is the loss value, is the output value of the enzymatic property prediction model, y is the true value of the enzymatic property, n is the number of samples, and cosh() is the hyperbolic cosine function.
[0014] Further, before training, it also includes preprocessing the microorganism features and enzymatic property features of the Daqu samples, and the preprocessing includes missing value processing, outlier processing, and normalization processing; after obtaining the normalized enzymatic property using the enzymatic property prediction model, the inverse processing corresponding to the normalization is adopted to restore the enzymatic property features to the true order of magnitude.
[0015] Further, during the training process, the logarithmic loss function, mean square error, or mean absolute error is used to calculate the error value of the enzymatic property prediction model, and when the error value is less than the preset value, the training of the enzymatic property prediction model is completed.
[0016] The present invention also provides an apparatus for evaluating the enzymatic properties and identifying key microorganisms of Daqu, which realizes the method for evaluating the enzymatic properties and identifying key microorganisms of Daqu as described above. The apparatus includes a data set acquisition module, a model training module, a microbial feature contribution module, a feature acquisition module, and a prediction and identification module. The data acquisition module is used to acquire the microbial characteristics and enzymatic property characteristics of Daqu samples. The model training module is used to train a neural network model to obtain an enzymatic property prediction model. The microbial feature contribution module is used to calculate the SHAP value of the microbial features for the prediction result, and evaluate the contribution degree of each microbial feature according to the calculated SHAP value, and select microorganisms as key microorganisms based on the contribution degree. The feature acquisition module is used to acquire the microbial characteristics of the Daqu to be predicted. The prediction and identification module is used to input the microbial characteristics of the Daqu to be predicted into the enzymatic property prediction model to obtain the prediction result of the enzymatic property characteristics of the Daqu, and use the microbial feature contribution module to obtain the key microorganism identification result.
[0017] Further, the microbial characteristics include the types and quantities of microbial populations during the Daqu fermentation process. The types of the microbial populations include Bacillus, Aspergillus, Saccharomyces cerevisiae, Pediococcus acidilactici, and Staphylococcus. The enzymatic property characteristics include saccharifying power, fermenting power, liquefying power, and esterifying power.
[0018] The beneficial effects of the present invention: The present invention provides a method and apparatus for evaluating the enzymatic properties and identifying key microorganisms of Daqu. By acquiring the microbial characteristics and enzymatic property characteristics of Daqu samples, training a neural network model to obtain an enzymatic property prediction model, constructing a microbial feature contribution module based on the SHAP theory to calculate the SHAP value of the microbial features for the prediction result, evaluating the contribution degree of each microbial feature according to the calculated SHAP value, selecting microorganisms as key microorganisms based on the contribution degree, using the enzymatic property prediction model to obtain the prediction result of the enzymatic property characteristics of the Daqu to be tested, and using the microbial feature contribution module to obtain the key microorganism identification result, a quantitative mapping relationship between the Daqu microbial community and the enzymatic properties is established, solving the problem that the influence of each microorganism on the enzymatic properties is not clear, which is beneficial to quality control in Daqu production, and provides a theoretical support and practical tool for improving the fermentation effect and optimizing the Daqu formula. Description of the Drawings
[0019] Figure 1 is the flow chart of the method for evaluating the enzymatic properties and identifying key microorganisms of Daqu provided by the present invention;
[0020] Figure 2 is the comparison chart of the predicted saccharifying power value and the true value output by the enzymatic property prediction model;
[0021] Figure 3It is a comparison chart of the predicted fermentation power value and the true value output by the enzymatic performance prediction model;
[0022] Figure 4 It is a comparison chart of the predicted liquefaction power value and the true value output by the enzymatic performance prediction model;
[0023] Figure 5 It is a comparison chart of the predicted esterification power value and the true value output by the enzymatic performance prediction model;
[0024] Figure 6 It is a bar chart of the average SHAP values of 15 microorganisms. Specific implementation manner
[0025] In view of the problem that the influence of various microorganisms in the existing Daqu on the enzymatic performance is not clear, the present invention provides a method for evaluating the enzymatic performance of Daqu and identifying key microorganisms, as Figure 1 shown, including the following steps:
[0026] S1. Obtain the microbial characteristics and enzymatic performance characteristics of the Daqu sample;
[0027] Specifically, the microbial characteristics include the types and quantities of microbial populations during the Daqu fermentation process, and the types of the microbial populations include Bacillus, Aspergillus, Saccharomyces cerevisiae, Pediococcus acidilactici, Staphylococcus, etc.; the enzymatic performance characteristics include saccharifying power, fermentation power, liquefaction power, esterification power, etc. Thus, a data set composed of microbial characteristics and enzymatic performance characteristics is obtained for training the neural network model.
[0028] S2. Establish a neural network model, use the microbial characteristics as the input, and the enzymatic performance characteristics as the output, train the neural network model to obtain an enzymatic performance prediction model;
[0029] Specifically, the neural network model includes one or a combination of MLP, LSTM, RNN, CNN, GRU, and Transformer. Before training, it also includes preprocessing the microbial characteristics and enzymatic performance characteristics of the Daqu sample. The preprocessing includes missing value processing, outlier processing, and normalization processing, so as to ensure the integrity and accuracy of the data and eliminate the influence of different microbial magnitudes; during training, the data set composed of microbial characteristics and enzymatic performance characteristics is divided into a training set and a validation set. The loss function of the enzymatic performance prediction model is: Among them, is the loss value, is the output value of the enzymatic performance prediction model, y is the true value of the enzymatic performance, n is the number of samples, and cosh() is the hyperbolic cosine function; the logarithmic loss function, mean square error, or mean absolute error is used to calculate the error value of the enzymatic performance prediction model. When the error value is less than the preset value, the training of the enzymatic performance prediction model is completed; after obtaining the normalized enzymatic performance using the enzymatic performance prediction model, the inverse process corresponding to the normalization is used to restore the enzymatic performance characteristics to the true order of magnitude.
[0030] S3. Construct a microbial feature contribution module based on the SHAP theory to calculate the SHAP value of the microbial features for the prediction result, and evaluate the contribution degree of each microbial feature according to the calculated SHAP value. Select the microorganisms as key microorganisms based on the contribution degree.
[0031] Specifically, the calculation formula of the SHAP value is: where, φ j is the SHAP value of the j-th microorganism; N is the set of all microbial features; N\{j} is the set of all subsets excluding the j-th microbial feature, S is a subset in N\{j}; v(S) is the output value of the enzymatic performance prediction model when only the subset S is included, and v(S∪{j}) - v(S) represents the change in the output value of the enzymatic performance prediction model after adding the j-th microbial feature; |S| is the size of the subset S; |N| is the total number of microbial features. When evaluating the contribution degree of each microbial feature, the average SHAP value of the microbial features for all enzymatic performance features is used. When selecting key microorganisms, select the top N microorganisms ranked from largest to smallest in terms of the average SHAP value, or microorganisms whose average SHAP value exceeds the threshold.
[0032] S4. Obtain the microbial features of the Daqu to be predicted, use the enzymatic performance prediction model to obtain the prediction result of the enzymatic performance features of the Daqu to be measured, and use the microbial feature contribution module to obtain the key microorganism recognition result.
[0033] The present invention also provides an apparatus for evaluating the enzymatic properties and identifying key microorganisms of Daqu, which realizes the method for evaluating the enzymatic properties and identifying key microorganisms of Daqu as described above. The apparatus includes a data set acquisition module, a model training module, a microbial feature contribution module, a feature acquisition module, and a prediction and identification module; the data acquisition module is used to acquire the microbial characteristics and enzymatic property characteristics of Daqu samples; the model training module is used to train a neural network model to obtain an enzymatic property prediction model; the microbial feature contribution module is used to calculate the SHAP values of microbial characteristics for the prediction results, and evaluate the contribution degree of each microbial characteristic according to the calculated SHAP values, and select microorganisms as key microorganisms based on the contribution degree; the feature acquisition module is used to acquire the microbial characteristics of the Daqu to be predicted; the prediction and identification module is used to input the microbial characteristics of the Daqu to be predicted into the enzymatic property prediction model to obtain the prediction result of the enzymatic property characteristics of the Daqu, and use the microbial feature contribution module to obtain the key microorganism identification result.
[0034] Specifically, the microbial characteristics include the types and quantities of microbial populations during the Daqu fermentation process. The types of microbial populations include Bacillus, Aspergillus, Saccharomyces cerevisiae, Pediococcus acidilactici, and Staphylococcus; the enzymatic property characteristics include saccharifying power, fermenting power, liquefying power, and esterifying power.
[0035] Taking 15 microorganisms of a certain Daqu and four enzymatic properties including saccharifying power, fermenting power, liquefying power, and esterifying power as examples, the neural network is trained to obtain the corresponding enzymatic property prediction model, and the four enzymatic properties are predicted. The comparison between the predicted saccharifying power value output by the enzymatic property prediction model and the true value is as Figure 2 shown. The comparison between the predicted fermenting power value output by the enzymatic property prediction model and the true value is as Figure 3 shown. The comparison between the predicted liquefying power value output by the enzymatic property prediction model and the true value is as Figure 4 shown. The comparison between the predicted esterifying power value output by the enzymatic property prediction model and the true value is as Figure 5 shown. The average SHAP values of 15 microorganisms with respect to the four enzymatic properties are as Figure 6 shown. Thus, it can be concluded that microorganisms 5, 12, 1, 2, and 11 are the top 5 microorganisms with the highest average SHAP values, and these 5 microorganisms are the key microorganisms.
Claims
1. Method for evaluating enzymatic properties of Daqu and identifying key microorganisms, characterized in that, It includes the following steps: S1. Obtain the microbial characteristics and enzymatic performance characteristics of the Daqu samples; S2. Establish a neural network model, using the microbial characteristics as the input and the enzymatic performance characteristics as the output, train the neural network model to obtain an enzymatic performance prediction model; S3. Calculate the SHAP values of the microbial characteristics for the prediction results based on the SHAP theory, and evaluate the contribution degree of each microbial characteristic according to the calculated SHAP values, and select the microorganisms as key microorganisms based on the contribution degree; S4. Obtain the microbial characteristics of the Daqu to be predicted, use the enzymatic performance prediction model to obtain the prediction result of the enzymatic performance characteristics of the Daqu to be measured, and use the microbial characteristic contribution module to obtain the key microorganism recognition result.
2. The method for evaluating the enzymatic properties of Daqu and identifying key microorganisms according to claim 1, characterized in that, The microbial characteristics include the types and quantities of microbial populations during the Daqu fermentation process, and the types of the microbial populations include Bacillus, Aspergillus, Saccharomyces cerevisiae, Pediococcus acidilactici, and Staphylococcus.
3. The method for evaluating the enzymatic properties and identifying key microorganisms of Daqu according to claim 1, characterized in that, The enzymatic performance characteristics include: saccharifying power, fermenting power, liquefying power, and esterifying power.
4. The method for evaluating the enzymatic properties of Daqu and identifying key microorganisms according to claim 1, characterized in that The neural network model includes one or a combination of more than one of MLP, LSTM, RNN, CNN, GRU, and Transformer.
5. The method for evaluating the enzymatic properties and identifying key microorganisms of Daqu according to claim 1, characterized in that, The calculation formula of SHAP value is as follows: where φ j is the SHAP value of the j-th microorganism; N is the set of all microorganism features; N\{j} is all subsets that do not contain the j-th microorganism feature, S is a subset in N\{j}; v(S) is the output value of the enzymatic property prediction model when only subset S is included, and v(S∪{j}) - v(S) represents the change in the output value of the enzymatic property prediction model brought about by adding the j-th microorganism feature; |S| is the size of subset S; |N| is the total number of microorganism features.
6. The method for evaluating the enzymatic properties of Daqu and identifying key microorganisms according to claim 1, characterized in that, The loss function of the enzymatic performance prediction model is as follows: where is the loss value, is the output value of the enzymatic performance prediction model, y is the true value of the enzymatic performance, n is the number of samples, and cosh() is the hyperbolic cosine function.
7. The method for evaluating the enzymatic properties and identifying key microorganisms of Daqu according to claim 1, characterized in that, Before training, it also includes preprocessing the microbial characteristics and enzymatic performance characteristics of the Daqu samples, and the preprocessing includes missing value processing, outlier processing, and normalization processing; after obtaining the normalized enzymatic performance using the enzymatic performance prediction model, perform the inverse processing corresponding to the normalization to restore the enzymatic performance characteristics to the true order of magnitude.
8. The method for evaluating the enzymatic properties of Daqu and identifying key microorganisms according to claim 1, characterized in that, During the training process, use the logarithmic loss function, mean square error, or mean absolute error to calculate the error value of the enzymatic performance prediction model. When the error value is less than the preset value, the training of the enzymatic performance prediction model is completed.
9. Device for evaluating the enzymatic properties of Daqu and identifying key microorganisms, characterized in that, Implement the method for evaluating the enzymatic performance and identifying key microorganisms of Daqu as described in claim 1. The device includes a data set acquisition module, a model training module, a microbial characteristic contribution module, a characteristic acquisition module, and a prediction and identification module; the data acquisition module is used to obtain the microbial characteristics and enzymatic performance characteristics of the Daqu samples; the model training module is used to train the neural network model to obtain an enzymatic performance prediction model; the microbial characteristic contribution module is used to calculate the SHAP values of the microbial characteristics for the prediction results, and evaluate the contribution degree of each microbial characteristic according to the calculated SHAP values, and select the microorganisms as key microorganisms based on the contribution degree; the characteristic acquisition module is used to obtain the microbial characteristics of the Daqu to be predicted; the prediction and identification module is used to input the microbial characteristics of the Daqu to be predicted into the enzymatic performance prediction model to obtain the prediction result of the enzymatic performance characteristics of the Daqu, and use the microbial characteristic contribution module to obtain the key microorganism recognition result.
10. The enzymatic property evaluation and key microorganism identification device for Daqu according to claim 9, characterized in that, The microbial characteristics include the types and quantities of microbial populations during the Daqu fermentation process, and the types of the microbial populations include Bacillus, Aspergillus, Saccharomyces cerevisiae, Pediococcus acidilactici, and Staphylococcus; The enzymatic performance characteristics include: saccharifying power, fermenting power, liquefying power, and esterifying power.