Intelligent brewing process of traditional modern fusion type Moungu Qinglian wine

Abstract

The invention discloses a traditional modern fusion type intelligent brewing process and system for Moungu Qinglian wine, and belongs to the technical field of white spirit brewing. According to the process, the accurate proportioning (25: 10: 35: 20: 10) of glutinous rice (amylopectin is greater than or equal to 80%), lotus seeds (starch is greater than or equal to 60%) and other raw materials is realized through an Internet of Things weighing system; the temperature and humidity (25 + / -2 DEG C / 75 + / -5%) of the cellar are dynamically regulated and controlled by adopting an LSTM model, and the rancidity rate is smaller than or equal to 1% by combining with caproic acid bacteria (larger than or equal to 106 CFU / g) enriched in the yellow water extract; quality-based liquor picking is realized based on near infrared spectrum and machine vision (the premium grade rate of middle-stage liquor is 92%); 12 flavor substances serve as input through a BP neural network, and a blending ratio (flavor similarity is larger than or equal to 98%) is output; and full-process tracing is realized by adopting a block chain technology. The wine yield is greater than or equal to 37.8% and is improved by 15.2% compared with that of a traditional process, the stability of ethyl hexanoate is improved by 68%, and the technical problems of inaccurate raw material ratio, uncontrollable fermentation, poor flavor stability and the like in the traditional process are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of liquor brewing, in particular to an intelligent brewing process for traditional and modern fusion Mugouqinglian liquor. BACKGROUND

[0002] As a traditional fermented beverage in China, the brewing process of liquor has a long history, but the traditional production mode has long relied on manual experience, which has the following technical defects:

[0003] 1. Inadequate raw material processing: the traditional process lacks precise detection means for key indicators such as waxy rice amylopectin and lotus seed starch, relying only on manual experience for proportioning, resulting in low utilization of raw materials and generally less than 35% of liquor yield.

[0004] 2. Uncontrollable fermentation process: pit temperature and humidity depend on natural environment adjustment, with large fluctuation range (±5℃), unstable microbial metabolism, high risk of rancidity (up to 12% for traditional process), and long fermentation cycle (180 days for double bottom fermentation), resulting in low production efficiency.

[0005] 3. Distillation and blending rely on experience: the traditional liquor cutting process relies on manual sensory judgment, with less than 70% of high-quality middle-liquor; the blending process lacks scientific and quantitative standards, with poor flavor consistency and more than 25% fluctuation in key ester content such as ethyl hexanoate.

[0006] 4. Difficult quality traceability: production data records are scattered, making it impossible to achieve full-process reliable traceability, affecting product standardization and brand credibility.

[0007] In recent years, some enterprises have tried to introduce automated equipment (such as temperature control sensors and online detectors), but still have the following problems:

[0008] 1. Single-point optimization, unable to achieve full-process intelligent collaborative control;

[0009] 2. Lack of dynamic control model, unable to adapt to nonlinear changes in the fermentation process;

[0010] 3. Serious data island phenomenon, insufficient application of blockchain and other traceability technologies.

[0011] The present application aims to solve the above problems by deeply integrating Internet of Things, AI algorithms and multi-spectral analysis technology to build an intelligent brewing system from raw materials to blending, improving liquor yield (38.5%) and reducing rancidity rate (≤1%), while achieving precise control of flavor substances (similarity ≥98%), providing an innovative solution for the upgrading of traditional liquor industry. SUMMARY

[0012] The technical problem to be solved by the present application is the problem of inaccurate raw material proportioning, uncontrollable fermentation process, poor flavor stability and difficult quality traceability in the traditional liquor brewing process.

[0013] The technical scheme adopted by the present application is a traditional and modern fusion type Mugou Qinglian liquor intelligent brewing process, comprising the following steps:

[0014] (1) Intelligent raw material pretreatment:

[0015] Waxy rice (amylopectin ≥ 80%), lotus seed (starch ≥ 60%), sorghum, rice, and wheat are proportioned at 25:10:35:20:10 by the Internet of Things weighing system;

[0016] After the raw materials are crushed, they are mixed with raw bran, and the steam condensate water from steaming bran is recycled as the water for measuring;

[0017] (2) Intelligent steaming and cooking:

[0018] Grain residues, red residues, and face residues are layered into a steaming pot, and a steam pressure sensor maintains the temperature at 105±2℃;

[0019] (3) AI fermentation management:

[0020] After the fermented grains are cooled to 25±1℃, the enriched bacteria (caproic acid bacteria ≥ 106 CFU / g, and yeast survival rate ≥ 95%) are inoculated;

[0021] The cellar intelligent temperature control system maintains 25±2℃ / humidity 75±5%, and the LSTM model dynamically regulates the fermentation period: 90 days for the conventional type and 180 days for the double wheel bottom type;

[0022] (4) Intelligent distillation and wine picking:

[0023] Based on near-infrared spectrum real-time detection of fermented grain components, the proportion of wine head (alcohol content ≥ 75%vol) is controlled at 3%, the middle section wine is controlled at 60% (60-70%vol), and the wine tail is controlled at 10% (≤40%vol);

[0024] Machine vision linkage multispectral detection (ethyl caproate ≥ 2.5g / L, ethyl acetate ≥ 1.2g / L) realizes quantity and quality wine picking;

[0025] (5) Digital aging and blending:

[0026] The base liquor is implanted with an RFID tag into a pottery jar library, and a BP neural network model takes 12 kinds of flavor substances as input and outputs the blending ratio (flavor similarity ≥ 98%);

[0027] The 12 kinds of flavor substances are ethyl caproate, ethyl acetate, ethyl lactate, ethyl butyrate, ethyl valerate, caproic acid, acetic acid, lactic acid, butyric acid, isoamyl alcohol, β-phenylethanol, and 2,3-butanediol;

[0028] According to the embodiment ratio: lotus seed wine 67% + double wheel wine 23.5% + seasoning old wine 9.5%.

[0029] (6) Whole-process quality control:

[0030] AI image recognition quality inspection (automatically rejected when transmittance ≤ 90% NTU, foreign matter recognition accuracy ≥ 99.5%);

[0031] A blockchain system using Hyperledger Fabric architecture, each node synchronizes data through gRPC protocol, and the block generation time interval is set to 10 seconds, realizing data traceability from raw material inspection to finished product (alcohol yield ≥ 37.8%, rancidity ≤ 1%).

[0032] As a further scheme of the present application: the lotus seed starch detection in step (1) adopts GB 5009.9-2016 standard, and waxy rice amylopectin is verified by iodine colorimetry.

[0033] As a further scheme of the present application: in step (3), the enrichment of caproic acid bacteria needs to add yellow water extract with pH = 4.0-4.5, and anaerobic culture at 30°C for 7 days.

[0034] As a further scheme of the present application: the preparation method of the yellow water extract is as follows: taking the pit bottom yellow water, filtering it through a 0.22 μm membrane, and then concentrating it to 20% of the original volume at 60°C in a rotary evaporator, the total acid content is measured to be 8.5±0.3 g / L (calculated as acetic acid).

[0035] As a further scheme of the present application: in step (4), the flavor substance control range of the middle section wine is: ethyl lactate 0.8-1.8 g / L, ethyl caproate 2.8-3.5 g / L.

[0036] As a further scheme of the present application: in step (5), the input layer of the BP neural network contains the base wine age and the concentration of 12 kinds of flavor substances detected by GC-MS, and the output layer gives the blending scheme of alcohol content 52±0.5%vol and total ester ≥4.5 g / L.

[0037] A traditional and modern fusion type Mugou Qinglian liquor intelligent brewing system, comprising the following mutually coordinated subsystems:

[0038] (1) Thermodynamic regulation subsystem

[0039] Based on the non-steady-state heat transfer equation, the temperature field of the pit is accurately controlled:

[0040]

[0041] Where, ρ = 1100 kg / m 3 : density of fermented grains (measured value); c p= 1.8 kJ / (kg·K): specific heat capacity of fermented grains (determined by differential scanning calorimetry); k = 0.48 W / (m·K): effective thermal conductivity of fermented grains; Q bio = βe -γt : heat release term of microbial fermentation (β = 85 W / m 3 , γ = 0.015 h -1 );

[0042] (2) Spectral detection subsystem

[0043] Multi-component synchronous detection is achieved by using improved Lambert-Beer law:

[0044]

[0045] Wherein, c i : concentration of the i-th flavor substance (such as ethyl caproate); ε i : characteristic absorption coefficient (ε of ethyl caproate = 280 L·mol -1 ·cm -1 ); l = 10 mm: optical path of flow detection cell; φ i (λ): wavelength selection function (Gaussian type, half-peak width 15 nm);

[0046] (3) Fluid dynamics subsystem

[0047] Steam flow in distillation process satisfies Navier-Stokes equation:

[0048]

[0049] Wherein, v: steam flow velocity field (m / s); p: pressure distribution in retort (Pa); μ = 1.2 × 10 -5 Pa·s: dynamic viscosity of steam; F: gravitational volume force (F = ρg);

[0050] (4) Chemical kinetics subsystem

[0051] The material conversion rate equation of fermentation process is:

[0052]

[0053] Wherein, k1 = 2.3 × 10 -4 L / (mol·h): esterification reaction rate constant; k2 = 5.6 × 10 -6 h -1 : ester hydrolysis rate constant;

[0054] (5) Control execution subsystem

[0055] Improved PID algorithm is used to realize process control:

[0056]

[0057] wherein, K p = 8.5: proportional gain; K i = 0.15: integral gain; K d = 1.2: derivative gain; η = 0.05: second order damping coefficient.

[0058] Advantages of the present application:

[0059] 1. Precise raw material ratio

[0060] The Internet of Things weighing system and standard detection methods (such as GB 5009.9-2016 lotus starch detection) are adopted to realize accurate proportioning (25:10:35:20:10) of raw materials such as waxy rice (amylopectin ≥ 80%) and lotus seeds (starch ≥ 60%), thereby guaranteeing consistency of fermentation substrates from the source and increasing the liquor yield to ≥ 37.8% (only 32.1% for traditional process).

[0061] 2. Intelligent regulation of fermentation process

[0062] Dynamic fermentation management based on LSTM model, combined with pit temperature control system (25±2℃ / humidity 75±5%) and enrichment of caproic acid bacteria (≥ 10 6 CFU / g) in yellow water extract, reduces the rancidity rate to ≤ 1% (12% for traditional process), shortens the double-wheel bottom fermentation period to 180 days, and increases the stability of flavor substance output by 68% (± 8% fluctuation of ethyl caproate vs. ± 25% of traditional).

[0063] 3. High precision of distillation and wine picking

[0064] The composition of fermented grains is detected in real time by near-infrared spectroscopy, and accurate interception is performed according to 3% (≥ 75% vol) of wine head, 60% (60-70% vol) of middle-stage wine, and 10% (≤ 40% vol) of wine tail, combined with machine vision quality and quantity wine picking (ethyl caproate ≥ 2.5 g / L, ethyl acetate ≥ 1.2 g / L), thereby increasing the superior rate of middle-stage wine from 68% to 92%.

[0065] 4. Scientific flavor blending

[0066] Based on the BP neural network model, 12 kinds of flavor substances (such as ethyl caproate 2.8-3.5 g / L, ethyl lactate 0.8-1.8 g / L) are taken as inputs, and the blending ratio (lotus wine 67% + double-wheel wine 23.5% + flavoring old wine 9.5%) is taken as output, with a flavor similarity of ≥ 98% and an alcohol content control accuracy of ± 0.5% vol.

[0067] 5. Quality traceability of the whole process

[0068] The liquor aging data is recorded by RFID tags by using the Hyperledger Fabric blockchain architecture (block generation interval 10 seconds), full-link traceability from raw material inspection to finished product delivery is realized, foreign matter identification accuracy is greater than or equal to 99.5%, the light transmittance exceeding the standard (greater than 90% NTU) is automatically removed, and the quality control efficiency is improved by more than 90%. BRIEF DESCRIPTION OF DRAWINGS

[0069] Figure 1 A liquor production process flowchart of the traditional and modern fusion type Mugou Qinglian liquor intelligent brewing process. DETAILED DESCRIPTION

[0070] The present application will be further described below.

[0071] Example 1: Standard intelligent brewing process (2023 October batch)

[0072] 1. Raw material pretreatment

[0073] Proportion: 25% waxy rice (amylopectin 82%), 10% lotus seed (starch 63%), 35% sorghum, 20% rice, 10% wheat.

[0074] Detection method: lotus seed starch is detected according to GB 5009.9-2016, and waxy rice amylopectin is verified by iodine colorimetry.

[0075] Cooking: steam pressure 0.15 MPa, temperature 105±2℃, cooking time 45 minutes.

[0076] 2. Fermentation management

[0077] Strain enrichment: add yellow water extract (total acid 8.5 g / L) with pH=4.3, anaerobic culture at 30℃ for 7 days, caproic acid bacteria concentration reaches 1.2×10 6 CFU / g.

[0078] Fermentation control: LSTM model dynamically adjusts temperature and humidity (25±1℃ / 75±3%), conventional fermentation for 90 days.

[0079] 3. Distillation and wine picking

[0080] Cutting ratio:

[0081] Wine head 3% (alcohol content 78%vol, ethyl caproate 5.8 g / L)

[0082] Middle section wine 60% (alcohol content 65%vol, ethyl caproate 3.2 g / L, ethyl lactate 1.5 g / L)

[0083] Wine tail 10% (alcohol content 38%vol)

[0084] 4. Blending and aging

[0085] Base liquor ratio: 2018 middle section lotus seed wine (35%) + 2018 early section lotus seed wine (32%) + 2017 double wheel wine (16%) + 2015 double wheel bottom cellar (7.5%) + 2012 seasoning wine (6.5%) + 2008 old wine (3%)

[0086] BP neural network output: alcohol content 52.1% vol, total esters 4.7 g / L, flavor similarity 98.3%.

[0087] 5. Quality testing

[0088] Transmittance: 88 NTU (qualified)

[0089] Foreign matter recognition: accuracy 99.6%

[0090] Yield: 38.5%

[0091] Example 2: Double wheel bottom intensified fermentation process (2023 December batch)

[0092] 1. Raw material adjustment

[0093] Lotus seed ratio increased: waxy rice 23%, lotus seed 12%, sorghum 35%, rice 20%, wheat 10%.

[0094] 2. Fermentation optimization

[0095] Double wheel bottom fermentation: extended to 180 days, cellar bottom hexanoic acid bacteria concentration reached 2.1 x 10 6 CFU / g.

[0096] Temperature control strategy: 25 ± 1°C for the first 60 days, 23 ± 1°C for the last 120 days (to promote ester synthesis).

[0097] 3. Distillation and wine extraction

[0098] Middle section wine flavor improved: ethyl hexanoate 3.5 g / L (upper limit value), ethyl lactate 1.8 g / L.

[0099] 4. Blending results

[0100] Base liquor ratio: 2019 double wheel wine (40%) + 2018 lotus seed wine (30%) + 2016 seasoning wine (20%) + 2010 old wine (10%)

[0101] Flavor indicators: total esters 5.1 g / L, ethyl hexanoate 3.4 g / L, consumer blind test preference rate increased to 89%.

[0102] 5. Comparative advantages

[0103] Yield: 37.2% (slightly lower than the standard process, but the flavor is richer).

[0104] Rancidity rate: 0.8% (due to low-temperature extended fermentation, microbial stability is higher).

[0105] Example 3: Low-cost rapid fermentation process (2024 March batch)

[0106] 1. Raw material simplification

[0107] Substitute part of sorghum with wheat: 25% waxy rice, 10% lotus seed, 30% sorghum, 20% rice, 15% wheat.

[0108] 2. Fermentation acceleration

[0109] Temperature increase: 28±2℃ (shorten the period to 60 days), but strict humidity control (70±5%) is required.

[0110] 3. Distillation adjustment

[0111] Reduced head cut: 2% (reduces high ester loss), middle section wine ratio increased to 65%.

[0112] 4. Blending scheme

[0113] Base liquor ratio: 2020 lotus seed wine (50%) + 2019 double wheel wine (30%) + 2018 seasoning wine (15%) + 2015 old wine (5%)

[0114] Flavor compromise: total esters 4.2g / L (slightly lower than standard), but production efficiency increased by 20%.

[0115] 5. Data comparison

[0116] Indicators Example 1 (standard) Example 2 (double wheel base) Example 3 (low cost) Fermentation cycle 90 days 180 days 60 days Yield of wine 38.5％ 37.2％ 36.8％ Total ester content 4.7 g / L 5.1 g / L 4.2 g / L Rancidity rate 1.0％ 0.8％ 1.5％ Production cost Benchmark +15％ -12％

[0117] Example comparison and analysis

[0118] 1. Balance of flavor and efficiency:

[0119] Example 1 (standard process) performs balanced in terms of liquor yield and flavor stability, suitable for mass production;

[0120] Example 2 (double wheel base) sacrifices part of the liquor yield, but the flavor levels are significantly improved, suitable for high-end products;

[0121] Example 3 (low cost) optimizes raw materials and fermentation, suitable for scenarios where market demand is large but cost-sensitive.

[0122] 2. Technical adaptability:

[0123] Double wheel base process requires a more stringent temperature control system;

[0124] Low-cost process can further optimize the rancidity rate by adjusting the raw material ratio.

[0125] 3. Consumer preference:

[0126] Flavor complexity is positively correlated with aging time (Example 2 has the highest blind preference rate);

[0127] The fast fermentation process needs to be compensated for the lack of flavor by blending (such as increasing the proportion of old wine).

[0128] Conclusion: The present application can be flexibly adapted to different production needs, and under the premise of ensuring the core indicators (liquor yield ≥ 37%, rancidity rate ≤ 1.5%), flavor customization and cost control are realized.

[0129] The above examples are only used to illustrate the technical solutions of the present application, but not to limit it; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that the technical solutions recorded in the foregoing examples can still be modified, or some technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A traditional-modern integrated intelligent brewing process for Mugou Qinglian liquor, characterized in that: Includes the following steps: (1) Intelligent raw material pretreatment: Glutinous rice (amylose ≥ 80%), lotus seeds (starch ≥ 60%), sorghum, rice, and wheat are weighed using an Internet of Things (IoT) system in a ratio of 25:10:35:20:

10. After the raw materials are crushed, they are mixed with raw rice bran, and the steam condensate from the rice bran is recycled as a water volume. (2) Intelligent steaming: Grain lees, red lees, and flour lees are layered and loaded into the steamer, and the steam pressure sensor maintains the temperature at 105±2℃. (3) AI fermentation management: After the mash is cooled to 25±1℃, it is inoculated with enriched bacterial strains (hexanoic acid bacteria ≥106 CFU / g, yeast survival rate ≥95%). The fermentation pit's intelligent temperature control system maintains a temperature of 25±2℃ and a humidity of 75±5%, while the LSTM model dynamically regulates the fermentation cycle: 90 days for conventional fermentation and 180 days for double-bottom fermentation. (4) Intelligent distillation and spirit extraction: Based on real-time detection of mash components using near-infrared spectroscopy, the proportions are controlled as follows: 3% of the first 3% of the distillate (alcohol content ≥75% vol), 60% of the middle 3% (60-70% vol), and 10% of the last 3% (≤40% vol). Machine vision-linked multispectral detection (ethyl hexanoate ≥ 2.5 g / L, ethyl acetate ≥ 1.2 g / L) enables quantitative and qualitative wine extraction; (5) Digital aging and blending: The base liquor is implanted with RFID tags and stored in ceramic jars. The BP neural network model takes the concentration of 12 flavor substances as input and outputs the blending ratio (flavor similarity ≥ 98%). The 12 flavor compounds are: ethyl hexanoate, ethyl acetate, ethyl lactate, ethyl butyrate, ethyl valerate, hexanoic acid, acetic acid, lactic acid, butyric acid, isoamyl alcohol, β-phenylethanol, and 2,3-butanediol. According to the example formula: 67% lotus seed wine + 23.5% double-boiled wine + 9.5% aged flavoring wine; (6) Full-process quality control: AI image recognition quality inspection (automatic rejection when light transmittance is ≤90% NTU, foreign object recognition accuracy ≥99.5%); The blockchain system, which adopts the Hyperledger Fabric architecture, synchronizes data among nodes via the gRPC protocol. The block generation interval is set to 10 seconds, enabling data traceability from raw material acceptance to finished product (yield ≥37.8%, spoilage rate ≤1%).

2. The intelligent brewing process for Mugou Qinglian wine, which integrates traditional and modern methods, as described in claim 1, is characterized in that... The lotus seed starch detection in step (1) adopts the GB 5009.9-2016 standard, and the glutinous rice amylopectin is verified by the iodine colorimetric method.

3. The intelligent brewing process for Mugou Qinglian wine, which integrates traditional and modern methods, as described in claim 1, is characterized in that... The enrichment of caproic acid bacteria in step (3) requires the addition of yellow water extract with pH=4.0-4.5 and anaerobic culture at 30℃ for 7 days.

4. The intelligent brewing process for Mugou Qinglian wine, which integrates traditional and modern methods, as described in claim 1, is characterized in that... The preparation method of the yellow water extract is as follows: Yellow water from the bottom of the cellar is filtered through a 0.22μm membrane and concentrated to 20% of its original volume in a rotary evaporator at 60℃. The total acid content is measured to be 8.5±0.3g / L (calculated as acetic acid).

5. The intelligent brewing process for Mugou Qinglian wine, which integrates traditional and modern methods, as described in claim 1, is characterized in that... The control range of flavor substances in the mid-stage wine mentioned in step (4) is: ethyl lactate 0.8-1.8 g / L, ethyl hexanoate 2.8-3.5 g / L.

6. The intelligent brewing process for Mugou Qinglian wine, which integrates traditional and modern methods, as described in claim 1, is characterized in that... The BP neural network input layer in step (5) includes the base wine vintage and the concentration of 12 flavor substances detected by GC-MS, and the output layer provides a blending scheme with an alcohol content of 52±0.5% vol and a total ester content of ≥4.5 g / L.

7. A traditional-modern integrated intelligent brewing system for Mugou Qinglian liquor, characterized in that: It includes the following mutually cooperating subsystems: (1) Thermodynamic control subsystem Precise control of the temperature field in the fermentation pit based on unsteady heat transfer equations: Where ρ = 1100 kg / m 3 : Bulk density of fermented mash (measured value); c p =1.8 kJ / (kg·K): Specific heat capacity of fermented grains (determined by differential scanning calorimetry); k = 0.48 W / (m·K): Effective thermal conductivity of fermented grains; Q bio =βe -γt Microbial fermentation exothermic term (β=85W / m) 3 ,γ=0.015h -1 ); (2) Spectral Detection Subsystem Simultaneous detection of multiple components is achieved using an improved Lambert-Beer law: Among them, c i ε: Concentration of the i-th flavor compound (e.g., ethyl hexanoate); i Characteristic absorption coefficient (ethyl hexanoate ε = 280 L·mol⁻¹) -1 ·cm -1 ); l = 10 mm: optical path length of the flow detection cell; φ i (λ): Wavelength selection function (Gaussian type, full width at half maximum 15 nm); (3) Fluid Dynamics Subsystem The vapor flow during distillation satisfies the Navier-Stokes equations: Where, v: steam velocity field (m / s); p: pressure distribution inside the steam pot (Pa); μ=1.2×10 -5 Pa·s: vapor dynamic viscosity; F: gravitational volume force (F=ρg); (4) Chemical Kinetics Subsystem The equation for the rate of material conversion during fermentation is: Where, k1 = 2.3 × 10 -4 L / (mol·h): esterification reaction rate constant; k2 = 5.6 × 10⁻⁶ -6 h -1 : Ester hydrolysis rate constant; (5) Control Execution Subsystem Process control is achieved using an improved PID algorithm: Among them, K p =8.5: Proportional gain; K i =0.15: Integral gain; K d =1.2: Differential gain; η=0.05: Second-order damping coefficient.

Images