Judgment method for stacking fermentation process of fermented grains and application of judgment method
By sampling and index detection of the wine mashed fermentation piles of liquor piles, a simple random forest model was established, and the fermentation process of liquor piles was simulated and judged, which solved the problem of inaccurate judgment in the existing technology and achieved higher judgment accuracy.
Patent Information
- Application Number
- CN202411704203.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-26
- Publication Date
- 2025-03-14
AI Technical Summary
The prior art cannot make a more accurate judgment on the fermentation process of liquor lees accumulation and fermentation, and it depends on the experience and subjective consciousness of production personnel.
By sampling the wine mashed with fermented piles of liquor, modeling samples are obtained, physical and chemical indicators are detected and microbial indicator changes are detected, suitable data are screened to establish a judgment model, and the fermentation process of the pile to be tested is simulated and judged using a simple random forest model.
A more accurate judgment of the accumulation and fermentation process of liquor lees is achieved, reducing subjectivity and improving the accuracy of judgment.
Smart Images
Figure CN119623277A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of wine brewing, and relates to a method for judging the fermentation process of fermented grains and an application thereof. Background Art
[0002] The brewing process of Maotai-flavor liquor requires about 5-10 days of pile fermentation first, and then about 30 days of anaerobic fermentation in the cellar before distillation. According to the traditional production process, the pile fermentation process is to place the pile on the open ground to achieve the enrichment and screening of the bacterial flora, and provide bacterial and enzyme system preparation for the subsequent cellar fermentation. As a brewing process for the accumulation of ethanol and flavor substances, anaerobic fermentation in the cellar is usually carried out in a cellar isolated from the environment, making it difficult to fully monitor and manually intervene in the fermentation process.
[0003] As an important process for shaping the microbial community structure system of Maotai-flavor liquor, pile fermentation is a key control stage in the production of Maotai-flavor liquor. As a solid-state fermentation process, pile fermentation has the characteristics of heterogeneous location and uneven fermentation process, that is, the fermentation process is affected by the shape of the pile, and different pile sites show different change trends. Therefore, in actual production, it is necessary to consider multiple indicators of multiple sites with different depths to comprehensively judge the maturity of pile fermentation.
[0004] However, the current judgment on the fermentation process of liquor lees still relies on the experience and subjective consciousness of production personnel, and it is impossible to make a more accurate judgment on the fermentation process of liquor lees. Summary of the invention
[0005] The present invention aims to provide a method for judging the fermentation progress of fermented grains in piles and its application, by sampling the fermented grains in a pile of liquor for fermentation to obtain a specific modeling sample, performing physical and chemical index detection and microbial index change detection on the modeling sample to obtain physical and chemical index data and microbial index data, and further screening out data suitable for modeling to establish a judgment model, and performing simulation judgment on the fermentation progress of the pile to be tested according to the established judgment model, so as to make a more accurate judgment on the fermentation progress of the liquor lees in piles in advance.
[0006] In a first aspect, the present invention provides a method for determining the progress of fermentation of fermented grains, comprising the following steps:
[0007] Obtaining modeling samples: sampling the fermented grains of liquor fermentation pile to obtain modeling samples;
[0008] Obtaining index data: performing physical and chemical index detection and microbial index change detection on the modeling sample to obtain physical and chemical index data and microbial index data;
[0009] Screening modeling data: screening the physical and chemical index data and the microbial index data as modeling data;
[0010] Establishing a judgment model: constructing a simplified random forest model based on the modeling data;
[0011] Judgment of the fermentation process: sampling the pile to be tested according to the method in the step of obtaining modeling samples to obtain the sample to be tested; detecting the physical and chemical indicators and microbial indicators of the sample to be tested according to the method in the step of obtaining indicator data; screening the variables required for the calculation of the simplified random forest model according to the step of screening modeling data; after obtaining the physical and chemical indicators and microbial indicator information data of the sample to be tested, importing them into the simplified random forest model to calculate the simulated fermentation process of the pile to be tested.
[0012] In some embodiments, the specific method of sampling is: sampling the fermented grains of the pile after 2 days, 4 days, 5 days, 7 days of stacking and when entering the cellar;
[0013] Preferably, sampling is performed at the core and surface of the pile to obtain core samples and surface samples, respectively.
[0014] In some embodiments, the physical and chemical index data include data information of the acidity, reducing sugar content, moisture, starch content, temperature and oxygen content of the fermented grains;
[0015] Preferably, the microbial indicator data includes data information on the bacterial and fungal microbial community structure of the mash.
[0016] In some embodiments, the bacteria include bacteria of the genera Lactobacillus and Ralstonia, and the fungus is a fungus of the genus Issatchenkia.
[0017] In some embodiments, the core sample and the surface sample form two simplified random forest models, namely the core model and the surface model. In the fermentation process judgment step, the core sample and the surface sample are respectively used as a sample to be tested. The physical and chemical indicators and microbial indicators of the samples to be tested are detected according to the method in the indicator data acquisition step. The variables required for the calculation of the simplified random forest model are screened according to the modeling data screening step. After the physical and chemical indicator and microbial indicator information data of the samples to be tested are obtained, they are introduced into the simplified random forest model to calculate the core and surface simulated fermentation processes of the pile to be tested.
[0018] In some embodiments, the microbial index data of the core model include bacteria of the genera Lactobacillus and Ralstonia of the fermented grains, and the physicochemical index data of the core model include acidity and reducing sugar content of the fermented grains;
[0019] Preferably, the microbial indicator data of the pile surface model are fungi of the genus Issatchenkia in the fermented mash, and the physical and chemical indicator data include reducing sugar, moisture and starch content of the fermented mash.
[0020] In some embodiments, in the step of screening modeling data, before screening the physical and chemical index data and the microbial index data as modeling data, the piles in the white wine fermentation piles that meet the cellaring standards are first screened as modeling piles, and then the physical and chemical index data and the microbial index data of the modeling piles are screened as modeling data.
[0021] In some embodiments, in the step of judging the fermentation progress, the pile to be tested is a pile that does not meet the cellaring standard.
[0022] In a second aspect, the present invention provides an application of any one of the above-mentioned judgment methods in regulating the stacking fermentation of mash.
[0023] In some embodiments, the method for regulating the fermentation of fermented grains is as follows: judging the fermentation progress of the pile to be tested according to any one of the judgment methods described above, and calculating the simulated fermentation progress of the pile core and the pile surface of the pile to be tested; respectively comparing the simulated fermentation progress of the pile core and the pile surface with the actual fermentation progress, and judging whether other process measures need to be taken for the pile to be tested according to the difference between the simulated fermentation progress of the pile core and the pile surface and the actual fermentation progress;
[0024] Preferably, if the simulated fermentation process of the pile core and the pile surface lags behind the actual fermentation process, a process measure of extending the fermentation time is taken;
[0025] Preferably, if the simulated fermentation process of the pile core and the pile surface meets the requirements for cellaring, the pile is cellared in advance.
[0026] In summary, the present application includes at least one of the following beneficial technical effects:
[0027] Specific modeling samples are obtained by sampling the mash from the liquor fermentation pile, and physical and chemical index tests and microbial index change tests are performed on the modeling samples to obtain physical and chemical index data and microbial index data, and the data suitable for modeling are further screened out to establish a judgment model, and the fermentation process of the pile to be tested is simulated and judged based on the established judgment model, so that a more accurate judgment can be made in advance on the fermentation process of the liquor lees accumulation. BRIEF DESCRIPTION OF THE DRAWINGS
[0028] Figure 1 This is a schematic diagram of the sampling location during the pile fermentation process;
[0029] Figure 2This is a diagram showing the succession of the first 20 microorganisms in the process of pile fermentation. Top: bacteria; bottom: fungi;
[0030] Figure 3 The probability diagram of the model's observed value being judged correct, where the left is the pile core model; the right is the pile surface model;
[0031] Figure 4 Build a variable importance ranking diagram for the model, where left: the core model; right: the table model. DETAILED DESCRIPTION
[0032] The technical solution of the present invention is further described below by specific embodiments, which do not limit the protection scope of the present invention. Some non-essential modifications and adjustments made by others based on the concept of the present invention still fall within the protection scope of the present invention.
[0033] Stack fermentation is an important stage in the production of Maotai-flavor liquor. Its fermentation quality has a direct effect on the formation of the fermentation microbial community structure in the cellar, which ultimately affects the quality and yield of the Maotai-flavor liquor base. However, the current judgment of the fermentation process still relies on the experience and subjective consciousness of the production staff, and fails to achieve a more accurate judgment of this process.
[0034] The present invention is the first to use a random forest model based on a dual-track system to judge the fermentation process of the core and surface of the pile of pile fermentation, and proposes a fermentation process discrimination method by comparing the actual fermentation process, the core fermentation process and the surface fermentation process. The present invention can effectively solve the problems of strong subjectivity and poor repeatability in the current judgment method, allowing production personnel to have a more comprehensive understanding of the state of pile fermentation. At the same time, it can also lay a solid foundation for the information reform of the pile fermentation process of Maotai-flavor liquor in the context of a closer integration of the traditional fermentation industry and information technology.
[0035] The technical scheme of the present invention is to obtain the calculation model of the core and surface of the standard fermentation pile by constructing a random forest model of the pile of standard fermentation days. The physical and chemical indexes and microbiological indexes of samples at different positions of the fermented mash to be tested are detected, and the fermentation progress of the core and surface of the pile is calculated by using the standard model. In actual production, the production situation and process measures of the pile to be tested can be determined by comparing the standard fermentation days of a normal pile.
[0036] In the process of model building, the applicant tried different physical and chemical indicators and microbial indicators as data for the construction of the core and surface models, and finally found that only when the core model used four indicators, namely Lactobacillus, Ralstonia, acidity and reducing sugar, and the surface model used four indicators, namely Issatchenkia, reducing sugar, moisture and starch content, the core model and surface model established could more accurately judge the fermentation process. When four indicators, namely Lactobacillus, Ralstonia, acidity and reducing sugar, were selected for the construction of the surface model, and four indicators, namely Issatchenkia, reducing sugar, moisture and starch content, were selected for the construction of the core model, the accuracy of the constructed model would be reduced due to the inconsistency of the fermentation characteristics of the core and surface of the pile to be tested. The top three indicators of importance, such as Issatchenkia, reducing sugar and moisture content, were selected for constructing the pile surface model, and the top three indicators of importance, such as Lactobacillus, Ralstonia and acidity, were selected for constructing the pile core model. The accuracy of the constructed simplified random forest model was not as good as that of the model of the present invention.
[0037] Example 1
[0038] A method for determining the fermentation progress of fermented grains comprises the following steps:
[0039] Step 1: Get modeling samples
[0040] The mash is sampled from the center and surface of the pile where the liquor is fermented to obtain the modeling samples.
[0041] Specifically, 10 piles of fermented Maotai-flavor liquor were selected as the tracking and monitoring objects. The pile center and pile surface positions (such as Figure 1 The fermented grains (as shown) were sampled, and 500 g of the fermented grains were sampled from the center and surface of the pile respectively.
[0042] Step 2: Get indicator data
[0043] The samples obtained in step 1 are tested for physical and chemical indicators and changes in microbial indicators to obtain physical and chemical indicator data and microbial indicator data.
[0044] Specifically, in this embodiment, the physical and chemical index data include the acidity, reducing sugar content, moisture, starch content, temperature, and oxygen content of the mash, and the microbial index data include data information on indicators such as the microbial community structure of bacteria and fungi.
[0045] Furthermore, the near-infrared method is used to detect the acidity, reducing sugar content, moisture and starch content of the mash. The specific detection method is: spread the sample evenly on the detection tray of the near-infrared instrument, push the detection tray into the near-infrared instrument, click the "Run" button in the integrated software, and the near-infrared instrument will automatically detect indicators such as the acidity, reducing sugar content, moisture and starch content in the mash.
[0046] The ethanol, acetic acid and lactic acid content of the fermented grains were detected by high performance liquid chromatography. The specific detection method is as follows: weigh 10g of fermented grains sample in a 250mL beaker, add 100mL of ethanol (95%) and soak for 2h, shake 3-4 times during the period, and perform a coarse filter with a qualitative filter paper. Then refrigerate the filtrate at -4℃ for 8h, filter it once with a 0.45μm microporous filter membrane, and then distill it under reduced pressure at 45℃ to 3-5L. Transfer the residual liquid to a 100mL volumetric flask with ultrapure water and make up the volume. Finally, dilute the sample 10 times and filter it through a 0.22μm filter membrane for injection.
[0047] The chromatographic conditions were Agilent Z0 RBXSB-C18 column (4.6 mm × 250 mm, 5 μm); column temperature 30 °C; mobile phase: methanol and 0.02 mol / L potassium dihydrogen phosphate solution (pH 2.1 adjusted with phosphoric acid) in a volume ratio of 15:85; flow rate 0.2 mL / min; detection wavelength 210 nm; injection volume 10 μL.
[0048] Electronic thermometers and residual oxygen meters are used to carry out in-situ detection of temperature and oxygen content. The specific detection method is: during the sampling process, insert the probes of the electronic thermometer and the residual oxygen meter into the sampling site, click the "Detect" button on the device, and the device will detect the temperature and oxygen of the mash at the detection site and record its detection values.
[0049] The microbial indicators use amplicon sequencing to detect the bacterial and fungal community structure. The specific detection method is: extract DNA from the mash sample, amplify the target region (such as 16S rRNA gene, ITS gene, etc.) through specific primers, purify the amplified product and perform high-throughput sequencing to obtain a large amount of sequence data, and import the data file into qiime2. After quality control, clustering, species annotation and other processing processes, the microbial community structure in the sample can be obtained.
[0050] Step 3: Filter modeling data
[0051] The piles that meet the cellar entry standards among the 10 tracking piles are selected as the modeling piles, and the physical and chemical indicators and microbiological indicators of the core and surface of the modeling piles are selected as modeling data.
[0052] Specifically, the present invention selects 8 piles that are determined by winemakers to meet the cellaring standards as modeling samples, and the other 2 piles that do not meet the cellaring standards are used as the piles to be tested.
[0053] Furthermore, the microorganisms were subjected to amplicon sequencing detection, and the top 20 microorganisms in the amplicon sequencing detection results were screened to draw a stacked bar graph to examine the abundance succession changes of various microorganisms (such as Figure 2 The results show that the Lactobacillus and Ralstonia genera in bacteria and the Issatchenkia genus in fungi have high abundance and large variation range in the accumulation process, which can be used as biomarker species of samples. Therefore, the above three microbial genera are selected as the modeling data of the samples. The modeling data information with 1 sample information (accumulation time), 9 physical and chemical indicators and 3 microbial indicators is shown in the following table.
[0054] List of indicators and their properties used to model data information tables
[0055]
[0056]
[0057] Step 4: Establish a judgment model
[0058] A simple random forest model was constructed based on the modeling data, and the performance of the model classifier was tested to screen the best modeling variable data, and the model was rebuilt to form two simple random forest models, the core model and the table model.
[0059] Specifically, the data information tables of the core and the heap table are imported into Rstudio respectively, the "heap duration" column is converted to a factor data type, and the train.forest = randomForest(duration~., data = data, importance = T, mtry = sqrt(ncol(data)-1)) function in the R language package randomForest is used to build a random forest model, and the probability that the observed value of the model is judged correct is examined ( Figure 3 ). Use importance_var = data.frame(importance(train.forest)) to evaluate the importance score of each variable in the model, and draw a graph to show the importance ranking of the variables ( Figure 4 ), identify key variables for model building.
[0060] Furthermore, based on the balance between the necessity of model construction and reducing the intensity of detection work, four key indicators were selected for the construction of the random forest model, namely, the core model adopted Lactobacillus, Ralstonia, acidity and reducing sugar, and the surface model adopted Issatchenkia, reducing sugar, moisture and starch content.
[0061] Select the selected indicators and use the train.forest_top = randomForest(duration~., data = data, importance = T, mtry = sqrt(ncol(data)-1)) function to rebuild the simplified random forest model of the core and the table.
[0062] Step 5: Determine the fermentation progress
[0063] The two piles that did not meet the cellaring standards described in the above-mentioned step of screening modeling data are selected as the piles to be tested. Samples are taken from the core and surface of the piles to be tested (test pile 1 and test pile 2) according to the method of step 1 to obtain the samples to be tested. The physical and chemical indicators and microbial indicators of the samples to be tested are tested according to the method of step 2. According to step 4, the variables required for calculating the simplified random forest model of the core and surface are screened to obtain the physical and chemical indicators and microbial indicator information data of the samples to be tested, as shown in the following table.
[0064] Test the physical and chemical indicators and microbiological indicators of the pile
[0065]
[0066]
[0067]
[0068] The physical and chemical indexes and microbiological indexes of the sample to be tested are imported into the discrimination model established in step 4 to calculate the simulated fermentation process of the pile center and the pile surface of the pile to be tested.
[0069] Specifically, the physical and chemical indexes and microbiological indexes of the test pile core and pile surface were imported into Rstudio, and the variables required for the calculation of the simplified random forest model were selected respectively. Based on the train.forest_top model, the test_predict=predict(train.forest_top,test_top) function was used to calculate the simulated fermentation process of the pile core and pile surface respectively. The results are shown in the following table:
[0070] Comparison of actual fermentation process and simulated fermentation process in the test pile
[0071] Heap to be tested Actual fermentation process (days) Simulated fermentation process in the core (days) Simulated fermentation process of pile surface (days) Test pile 1 2 2 2 Test pile 1 4 5 5 Test pile 1 6 7 7 Test pile 1 7 8 8 Test stack 2 2 2 2 Test stack 2 4 7 8 Test stack 2 6 8 8 Test stack 2 7 8 8
[0072] When both the core fermentation process and the surface fermentation process in the model reach the 8th day, it means that the actual fermentation process of the test pile meets the requirements for entering the cellar. By comparing the actual fermentation process of the test pile with the simulated fermentation process, it can be found that the indicators of test pile 1 and test pile 2 in actual production change very quickly. Test pile 1 has met the model's requirements for entering the cellar at 7 days, and test pile 2 has even reached the cellar standard at 6 days.
[0073] Example 2
[0074] A method for determining the progress of fermentation of fermented grains is applied to regulating fermentation of fermented grains, specifically:
[0075] According to the method in Example 1, the fermentation progress of the pile to be tested (test piles 3-5) is judged, and the simulated fermentation progress of the core and the surface of the pile to be tested is calculated. The simulated fermentation progress of the core and the surface of the pile are respectively compared with the actual fermentation progress, and it is judged whether other process measures are needed for the pile to be tested according to the difference in the process between the two. Specifically: if the model judges that the fermentation progress of the core and the surface of the pile lags significantly behind the actual fermentation progress, then the process measure of extending the fermentation time is taken; if the model judges that the fermentation progress of the core and the surface of the pile has reached the requirements for cellaring, then the pile is cellared in advance. The results are shown in the following table:
[0076] Comparison of actual fermentation process and simulated fermentation process in the test pile
[0077] Heap to be tested Actual fermentation process (days) Simulated fermentation process in the core (days) Simulated fermentation process of pile surface (days) Test stack 3 4 4 5 Test stack 4 7 5 5 Test pile 5 7 8 8
[0078] As shown in the table below, although the fermentation process on the surface of test pile 3 is slightly faster, it is consistent with the actual fermentation process of the pile as a whole. Therefore, based on the model, it is judged that the pile is fermenting normally. The fermentation process in the core of test pile 4 lags significantly behind the actual fermentation and surface fermentation processes. It is speculated that the fermentation delay may be caused by the influence of temperature on the stacking process. Therefore, based on the model, it may be necessary to take process measures to extend the fermentation time. The actual fermentation process of test pile 5 is only 7 days, but the model judges that both the core and surface fermentation processes have met the requirements for cellaring, so it can be cellared in advance.
[0079] Pile fermentation problems and technological measures based on model judgment
[0080] Heap Speculated reasons Process measures Test stack 3 Fermentation is normal No other process measures are required Test stack 4 Fermentation delay Prolong fermentation time Test pile 5 Fermentation in advance Early cellaring
[0081] Comparative Example 1
[0082] A method for judging the fermentation progress of fermented grains pile, which differs from Example 1 in that, in the step of establishing a discriminant model, four indicators, namely Lactobacillus, Ralstonia, acidity and reducing sugar, are selected for constructing a pile surface model, and four indicators, namely Issatchenkia, reducing sugar, moisture and starch content, are selected for constructing a pile core model. The constructed simplified random forest model judges the fermentation progress of fermented grains at each stage of the pile to be tested, the test pile 1 and the test pile 2, as shown in the following table:
[0083] Comparison of actual fermentation process and simulated fermentation process in the test pile
[0084] Heap Actual fermentation process (days) Simulated fermentation process in the core (days) Simulated fermentation process of pile surface (days) Test pile 1 2 2 2 Test pile 1 4 2 6 Test pile 1 6 4 8 Test pile 1 7 6 8 Test stack 2 2 2 2 Test stack 2 4 2 6 Test stack 2 6 4 4 Test stack 2 7 4 6
[0085] It can be found that by comparing the simulated fermentation process with the actual fermentation process, the accuracy of the index pairing method can be judged. The simulated fermentation process of the core and the simulated fermentation process of the heap surface have a large deviation, and are inconsistent with the actual fermentation process, so it is determined that the model is misjudged. It shows that due to the inconsistent fermentation characteristics of the core and the heap surface of the heap to be tested, 4 indicators such as Lactobacillus, Ralstonia, acidity and reducing sugar are selected for the construction of the heap surface model, and 4 indicators such as Issatchenkia, reducing sugar, moisture and starch content are used for the construction of the core model, which will cause the model to be misjudged, reduce the accuracy of the model, especially in the later stage of heap fermentation, and there is a large deviation between the actual fermentation progress. Therefore, based on the comparison result with the aforementioned embodiment 1, it is shown that the construction of the core model based on 4 indicators such as Lactobacillus, Ralstonia, acidity and reducing sugar, and the construction of the heap surface model based on 4 indicators such as Issatchenkia, reducing sugar, moisture and starch content, is the optimal combination for building the judgment of the heap fermentation process of Maotai-flavor liquor.
[0086] Comparative Example 2
[0087] A method for judging the fermentation progress of fermented grains pile, which differs from Example 1 in that, in the step of establishing a discriminant model, the top three indicators of importance, such as Issatchenkia, reducing sugar and moisture content, are selected for the construction of a pile surface model, and the top three indicators of importance, such as Lactobacillus, Ralstonia and acidity, are selected for the construction of a pile core model. The constructed simplified random forest model judges the fermentation progress of fermented grains at each stage of the test pile 1 and the test pile 2 as shown in the following table:
[0088] Comparison of actual fermentation process and simulated fermentation process in the test pile
[0089] Heap Actual fermentation process (days) Simulated fermentation process in the core (days) Simulated fermentation process of pile surface (days) Test pile 1 2 2 3 Test pile 1 4 5 5 Test pile 1 6 5 7 Test pile 1 7 6 8 Test stack 2 2 2 2 Test stack 2 4 7 6 Test stack 2 6 8 8 Test stack 2 7 6 8
[0090] The accuracy of the indicator pairing method can be judged by comparing the simulated fermentation process with the actual fermentation process. It can be found that although the simple model constructed using the top three indicators ranked by variable importance can still judge the stacking fermentation process as a whole, its accuracy is still lower than that of the model constructed with four indicators, indicating that further streamlining of indicators will result in a decrease in model accuracy.
[0091] It is to be understood that the present invention is described by some embodiments, and it is known to those skilled in the art that various changes or equivalent substitutions may be made to these features and embodiments without departing from the spirit and scope of the present invention. In addition, under the teachings of the present invention, these features and embodiments may be modified to adapt to specific circumstances and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the scope of protection of the present invention.
Claims
1. A method for determining the progress of fermentation of fermented grains, characterized in that: The following steps are involved: Obtaining modeling samples: sampling the fermented grains of liquor fermentation pile to obtain modeling samples; Obtaining index data: performing physical and chemical index detection and microbial index change detection on the modeling sample to obtain physical and chemical index data and microbial index data; Screening modeling data: screening the physical and chemical index data and the microbial index data as modeling data; Establishing a judgment model: constructing a simplified random forest model based on the modeling data; Determination of the fermentation progress: sampling the pile to be tested according to the method in the step of obtaining modeling samples to obtain the sample to be tested; detecting the physical and chemical indicators and microbial indicators of the sample to be tested according to the method in the step of obtaining indicator data; screening the variables required for the calculation of the simplified random forest model according to the step of screening modeling data; obtaining the physical and chemical indicators and microbial indicator information data of the sample to be tested and then importing them into the simplified random forest model to calculate the simulated fermentation progress of the pile to be tested.
2. The determination method according to claim 1, characterized in that: In the step of obtaining the modeling sample, the specific method of sampling is: sampling the fermented grains of the pile after 2 days, 4 days, 5 days, 7 days of stacking and when entering the cellar; Preferably, sampling is performed at the core and surface of the pile to obtain core samples and surface samples, respectively.
3. The determination method according to claim 1, characterized in that: The physical and chemical index data include data information of the acidity, reducing sugar content, moisture, starch content, temperature and oxygen content of the fermented grains; Preferably, the microbial indicator data includes data information on the bacterial and fungal microbial community structure of the mash.
4. The determination method according to claim 3, characterized in that: The bacteria include bacteria of the genera Lactobacillus and Ralstonia, and the fungi are fungi of the genus Issatchenkia.
5. The determination method according to claim 2, characterized in that: The core sample and the pile surface sample form two simplified random forest models, namely the core model and the pile surface model. In the fermentation process judgment step, the core sample and the pile surface sample are respectively used as a sample to be tested. The physical and chemical indicators and microbial indicators of the samples to be tested are tested according to the method in the indicator data acquisition step. The variables required for the calculation of the simplified random forest model are screened according to the modeling data screening step. After the physical and chemical indicator and microbial indicator information data of the samples to be tested are obtained, they are introduced into the simplified random forest model to calculate the core and surface simulated fermentation processes of the pile to be tested.
6. The determination method according to claim 5, characterized in that: The microbial index data of the core model include bacteria of the genera Lactobacillus and Ralstonia of the fermented grains, and the physical and chemical index data of the core model include acidity and reducing sugar content of the fermented grains; Preferably, the microbial indicator data of the pile surface model are fungi of the genus Issatchenkia in the fermented mash, and the physical and chemical indicator data include reducing sugar, moisture and starch content of the fermented mash.
7. The determination method according to claim 1, characterized in that: In the step of screening modeling data, before screening the physical and chemical index data and the microbial index data as modeling data, the piles in the liquor fermentation piles that meet the cellar entry standards are first screened as modeling piles, and then the physical and chemical index data and the microbial index data of the modeling piles are screened as modeling data.
8. The determination method according to any one of claims 1 to 7, characterized in that: In the step of judging the fermentation process, the pile to be tested is a pile that does not meet the cellaring standard.
9. Application of the judgment method as described in any one of claims 1 to 8 in regulating the stacking fermentation of fermented grains.
10. The use according to claim 9, characterized in that The method for regulating the fermentation of fermented grains is as follows: judging the fermentation process of the pile to be tested according to the judgment method described in any one of claims 1 to 8, and calculating the simulated fermentation process of the pile core and the pile surface of the pile to be tested; respectively comparing the simulated fermentation process of the pile core and the pile surface with the actual fermentation process, and judging whether other process measures are needed for the pile to be tested according to the difference between the simulated fermentation process of the pile core and the pile surface and the actual fermentation process; Preferably, if the simulated fermentation process of the pile core and the pile surface lags behind the actual fermentation process, a process measure of extending the fermentation time is taken; Preferably, if the simulated fermentation process of the pile core and the pile surface meets the requirements for cellaring, the pile is cellared in advance.