Real-time analysis system for wine liquid components, construction method and dynamic wine picking method

Abstract

This application provides a real-time analysis system for wine components, its construction method, and a dynamic wine extraction method, relating to the field of wine monitoring technology. The construction method of the real-time wine component analysis system includes: constructing base wine grading indicators; adjusting a near-infrared spectroscopy acquisition device by modifying its various parameters; acquiring representative samples, and based on standard detection methods, detecting various indicators on the representative samples to obtain standard physicochemical data, which is then used as a reference standard for the infrared spectroscopy acquisition device; using the adjusted near-infrared spectrum to detect various indicators on the representative samples, and correlating and matching the detection results with the standard physicochemical data. The wine component real-time analysis system construction method provided in this application enables rapid, accurate, and real-time analysis of wine components.

Description

Technical Field

[0001] This application relates to the field of signal wine monitoring technology, specifically to a real-time wine component analysis system, its construction method, and a dynamic wine extraction method. Background Technology

[0002] Currently, the traditional method of distillation in the baijiu industry is to "observe the foam" and "discard based on taste." This model relies entirely on the personal experience of master distillers to carry out the distillation operation—by observing the shape, size, and duration of the foam generated by the impact of the liquor, and combining their own sense of taste and smell to judge the alcohol content and flavor level of the liquor, they then divide the liquor into heads, middle sections, and tails, and determine the start and end points and segmentation standards for distillation.

[0003] The existing technology has the following problems:

[0004] (a) Skills transmission: Traditional skills rely on personal experience and face the risk of being lost.

[0005] Traditional manual winemaking techniques rely entirely on personal experience, with subjective and unquantifiable judgment criteria, making it difficult to form a replicable inheritance system. Senior winemaking masters are scarce and facing an aging population, making it difficult for young practitioners to quickly master core skills, resulting in a significant risk of the skills being lost to time.

[0006] (II) Talent Development: The training cycle for excellent wine tasters is long, and the supply of talent is insufficient.

[0007] Existing wine-picking models all heavily rely on excellent wine tasters, whose training cycle can take several to several decades. Furthermore, they lack systematic and quantitative standards and depend entirely on individual understanding and mentorship. This has led to a severe shortage of wine tasters in the industry, which restricts the stability of wine-picking quality and the implementation of intelligent technologies. Small and medium-sized enterprises are more significantly affected.

[0008] (III) Quality fluctuation: Human judgment is highly subjective, and the problem of uneven quality between batches of wine is prominent.

[0009] Traditional manual wine picking is subject to significant judgment errors due to individual differences among wine tasters, environmental factors, and physical conditions. This results in inaccurate wine picking segments and noticeable fluctuations in batch quality. The error rate of traditional manual wine picking is as high as 15%-20%, leading to a low yield of high-quality wine.

[0010] (iv) Efficiency bottleneck: High operational intensity, making it difficult to achieve large-scale replication.

[0011] Traditional manual distillation is labor-intensive and cannot be interrupted. Each person can only process a limited number of batches per day, making it difficult to adapt to large-scale production. In particular, the efficiency of multi-round distillation of Maotai-flavor liquor is a prominent shortcoming, with the extraction rate of premium liquor being less than 70%, which restricts the increase in production capacity.

[0012] (v) Data black box level: lack of quantitative data makes it difficult to build quality profiles.

[0013] Traditional manual winemaking decisions rely on experience, and the basis for judgment cannot be quantified or recorded, forming a "data black box". Summary of the Invention

[0014] The present application provides a method for constructing a real-time analysis system for wine components and a dynamic wine extraction method, which can effectively solve the above-mentioned problems.

[0015] The specific technical solution of this embodiment is as follows:

[0016] In a first aspect, embodiments of this application provide a method for constructing a real-time analysis system for wine components, including:

[0017] S10. Construct a base wine grading index, wherein each index corresponds to a flavor parameter set, wherein the flavor parameter set includes the grading threshold range of at least one of alcohol content, total acid, total esters, acetic acid, lactic acid, ethyl acetate, and ethyl lactate.

[0018] S20. Debug the near-infrared spectral acquisition device and adjust the various parameters of the near-infrared spectral acquisition device;

[0019] S30. Obtain representative samples, and based on standard detection methods, detect various indicators of the representative samples to obtain standard physicochemical data of the representative samples, and use the standard physicochemical data as a reference standard for application in infrared spectroscopy acquisition devices.

[0020] S40. The adjusted near-infrared spectrum is used to detect various indicators of the representative sample, and the detection results are correlated and matched with the standard physicochemical data. The correlated and matched data is sent to the iterative model for iteration to obtain the optimized parameters of the near-infrared spectral acquisition device.

[0021] In some embodiments, detecting various parameters of a representative sample using adjusted near-infrared spectroscopy includes the following steps:

[0022] S401, ensuring that the representative sample to be tested each time is not less than 80-120ml;

[0023] S402. Keep the representative sample warm so that the temperature of the representative sample is consistent with the actual wine receiving temperature.

[0024] S403. Perform tests on representative samples for various indicators.

[0025] In some embodiments, obtaining a representative sample includes the following steps:

[0026] S301. Collect samples from each receiving point in the winery, covering different production shifts, different batches, and different cellars. Samples are taken from each receiving point in three stages: the beginning, middle, and end of the wine.

[0027] S302. Screen all sample spectra and extract representative samples.

[0028] In some embodiments, the number of samples collected from each wine inlet in the winery is no less than 300, and the number of samples after spectral screening is no less than 100.

[0029] In some embodiments, after collecting samples from each inlet within the winery and before screening all sample spectra, the following steps are also included:

[0030] The samples should be sealed and stored at a temperature of 20-25℃.

[0031] In some embodiments, the iterative model includes:

[0032] The flavor quality calculation unit calculates the flavor quality index and the deviation of each flavor parameter based on the flavor parameter set.

[0033] (1)

[0034] Where F(t) is the flavor quality index, Xi is the i-th flavor parameter, and d Xi Let be the deviation of the i-th flavor parameter at time t, and wi be the weighting coefficient of the i-th flavor parameter. ;

[0035] (2)

[0036] (3)

[0037] (4)

[0038] in, and These represent the lower and upper limits of the ideal concentration of parameter X at alcohol content A, respectively.

[0039] The correction unit calculates the correction magnitude based on the difference between the flavor quality index and the historical average flavor quality index, and outputs a comprehensive quality score:

[0040] (5)

[0041] in, The correction magnitude is α, and the correction coefficient is α. This represents the historical average flavor quality index at alcohol content A.

[0042] (6)

[0043] in, It is a comprehensive quality score. It is a theoretical quality grade based on alcohol content;

[0044] The wine-picking decision unit dynamically adjusts the wine-picking segment points based on the time series of the comprehensive quality score, and outputs the corrected segment time points and the quality level of each segment.

[0045] The iterative update unit updates the weighting coefficients and ideal concentration ranges of flavor parameters based on the segmented deviation data.

[0046] In some embodiments, the iterative update unit includes:

[0047] The feedback collection subunit obtains the actual decision data of the wine picker on the segmentation point of this batch of wine picking, including the actual segmentation time point t1;

[0048] Deviation calculation sub-unit, calculates segmented deviation:

[0049] (7)

[0050] Where t2 represents the segmented time points of the model's predicted output;

[0051] The weight update sub-unit updates the weight coefficients based on the segmentation deviation and the deviation of each flavor parameter at the deviation time point:

[0052] (8)

[0053] in, It is the weight coefficient of the i-th flavor parameter after this iteration update. It is the weighting coefficient of the i-th flavor parameter used in the current batch. This is the learning rate, with a value ranging from 0.05 to 0.1. It is the time when the i-th flavor parameter deviates. The degree of deviation It indicates the time and location at which the segmented deviation occurs.

[0054] Secondly, embodiments of this application provide a real-time analysis system for wine components, which is constructed based on the construction method of the real-time analysis system for wine components according to any one of the second aspects.

[0055] Thirdly, embodiments of this application provide a dynamic wine extraction method for real-time analysis of wine components, including:

[0056] T10. Real-time spectral acquisition of the entire process of base liquor extraction;

[0057] T20. The collected data is sent to the real-time analysis system of wine components in the above embodiment to obtain the wine extraction results.

[0058] Compared with the prior art, the embodiments of this application have the following beneficial effects:

[0059] The method for constructing a real-time wine component analysis system provided in this application clarifies the specific composition and grading threshold range of the flavor parameter set by constructing base wine grading indicators, providing a clear judgment standard for subsequent wine component analysis. The near-infrared spectroscopy acquisition device is debugged and its parameters are adjusted to ensure the stability and accuracy of spectral acquisition, laying a hardware foundation for subsequent detection. Representative samples are acquired and standard physicochemical data are obtained based on standard detection methods, serving as a reference benchmark, providing a reliable basis for the correlation and matching of near-infrared spectral detection results. The adjusted near-infrared spectrum is used to detect representative samples, and the detection results are correlated and matched with the standard physicochemical data before being sent to an iterative model for iteration. This continuously optimizes the parameters of the near-infrared spectroscopy acquisition device, thereby improving its accuracy and reliability in wine component detection. Through continuous optimization of the iterative model, the final constructed real-time wine component analysis system can achieve rapid, accurate, and real-time analysis of wine components, providing strong technical support for applications such as dynamic wine harvesting. Attached Figure Description

[0060] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0061] Figure 1 This is a flowchart illustrating the construction method of a real-time analysis system for wine components provided in some embodiments of this application. Detailed Implementation

[0062] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0063] In the description of this application, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships based on the orientation or positional relationships shown in the accompanying drawings, are used only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "a plurality of" means two or more, unless otherwise explicitly specified.

[0064] The use of "applies to" or "configured to" in this application implies open and inclusive language, which does not exclude the applicability to or configuration to devices performing additional tasks or steps. Additionally, the use of "based on" implies openness and inclusivity, because processes, steps, calculations, or other actions "based on" one or more of the stated conditions or values ​​may in practice be based on additional conditions or values ​​beyond those stated.

[0065] In this application, the term "exemplary" is used to mean "used as an example, illustration, or description." Any embodiment described as "exemplary" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use this application. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that this application can be made without using these specific details. In other instances, well-known structures and processes are not described in detail to avoid obscuring the description of this application with unnecessary detail. Therefore, this application is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0066] Firstly, please refer to Figure 1 This application provides a method for constructing a real-time analysis system for wine components, including:

[0067] S10. Construct a base wine grading index, wherein each index corresponds to a flavor parameter set, wherein the flavor parameter set includes the grading threshold range of at least one of alcohol content, total acid, total esters, acetic acid, lactic acid, ethyl acetate, and ethyl lactate.

[0068] S20. Debug the near-infrared spectral acquisition device and adjust the various parameters of the near-infrared spectral acquisition device;

[0069] S30. Obtain representative samples, and based on standard detection methods, detect various indicators of the representative samples to obtain standard physicochemical data of the representative samples, and use the standard physicochemical data as a reference standard for application in infrared spectroscopy acquisition devices.

[0070] S40. The adjusted near-infrared spectrum is used to detect various indicators of the representative sample, and the detection results are correlated and matched with the standard physicochemical data. The correlated and matched data is sent to the iterative model for iteration to obtain the optimized parameters of the near-infrared spectral acquisition device.

[0071] Through the above embodiments, by constructing base wine grading indicators, the specific composition of the flavor parameter set and the grading threshold range are clarified, providing a clear judgment standard for subsequent wine component analysis. The near-infrared spectroscopy acquisition device was debugged and its parameters adjusted to ensure the stability and accuracy of spectral acquisition, laying the hardware foundation for subsequent detection. Representative samples were acquired and standard physicochemical data were obtained based on standard detection methods, serving as a reference benchmark, thus providing a reliable basis for the correlation and matching of near-infrared spectral detection results. The adjusted near-infrared spectra were used to detect representative samples, and the detection results were correlated and matched with the standard physicochemical data before being sent to an iterative model for iteration. This continuously optimizes the parameters of the near-infrared spectroscopy acquisition device, thereby improving its accuracy and reliability in wine component detection. Through continuous optimization of the iterative model, the final constructed real-time wine component analysis system enables rapid, accurate, and real-time analysis of wine components, providing strong technical support for applications such as dynamic wine harvesting.

[0072] In some embodiments, detecting various parameters of a representative sample using adjusted near-infrared spectroscopy includes the following steps:

[0073] S401, ensuring that the representative sample to be tested each time is not less than 80-120ml;

[0074] S402. Keep the representative sample warm so that the temperature of the representative sample is consistent with the actual wine receiving temperature.

[0075] S403. Perform tests on representative samples for various indicators.

[0076] By setting up the above embodiments, representative samples are kept at a constant temperature to match the actual wine receiving temperature. This insulation effectively reduces the interference of temperature fluctuations on spectral detection, improves the consistency between the detection environment and the actual production environment, and provides a more reliable spectral data foundation for subsequent data correlation matching and model iteration.

[0077] In some embodiments, obtaining a representative sample includes the following steps:

[0078] S301. Collect samples from each receiving point in the winery, covering different production shifts, different batches, and different cellars. Samples are taken from each receiving point in three stages: the beginning, middle, and end of the wine.

[0079] S302. Screen all sample spectra and extract representative samples.

[0080] By taking samples in stages from different production shifts, batches, and fermentation pits using the above-described embodiments, the various changes in the base liquor during the production process can be better covered, ensuring the broad representativeness of the collected samples. Furthermore, screening all sample spectra and extracting representative samples further guarantees the data quality used for subsequent standard physicochemical data testing and model training, reducing the interference of abnormal or atypical samples on the model's accuracy. This allows the constructed real-time liquor component analysis system to better adapt to the complex and ever-changing liquor conditions in actual production.

[0081] In some embodiments, the number of samples collected from each wine inlet in the winery is no less than 300, and the number of samples after spectral screening is no less than 100.

[0082] By setting the above embodiments to collect no fewer than 300 samples, the diversity and coverage of the samples can be guaranteed from a statistical perspective. This reduces model training bias caused by insufficient sample size and ensures sufficient data support for subsequent standard physicochemical data detection and model training. Furthermore, the number of samples after spectral screening is no fewer than 100, providing a sufficient number of effective samples for the model while ensuring data quality. This allows the model to learn the correlation between the spectral characteristics and physicochemical indicators of different types and characteristics of wines, thereby improving the model's generalization ability and prediction accuracy. This lays a data foundation for building a high-precision real-time wine component analysis system.

[0083] In some embodiments, after collecting samples from each inlet within the winery and before screening all sample spectra, the following steps are also included:

[0084] The samples should be sealed and stored at a temperature of 20-25℃.

[0085] By setting up the above embodiments, sealing and storing the sample at a temperature of 20-25°C can effectively reduce the escape of volatile flavor substances from the sample and the intrusion of external pollutants, ensuring that the physicochemical properties of the sample remain stable before testing.

[0086] In some embodiments, the iterative model includes:

[0087] The flavor quality calculation unit calculates the flavor quality index and the deviation of each flavor parameter based on the flavor parameter set.

[0088] (1)

[0089] Where F(t) is the flavor quality index, Xi is the i-th flavor parameter, and d Xi Let be the deviation of the i-th flavor parameter at time t, and wi be the weighting coefficient of the i-th flavor parameter. ;

[0090] (2)

[0091] (3)

[0092] (4)

[0093] in, and These represent the lower and upper limits of the ideal concentration of parameter X at alcohol content A, respectively.

[0094] The correction unit calculates the correction magnitude based on the difference between the flavor quality index and the historical average flavor quality index, and outputs a comprehensive quality score:

[0095] (5)

[0096] in, The correction magnitude is α, and the correction coefficient is α. This represents the historical average flavor quality index at alcohol content A.

[0097] (6)

[0098] in, It is a comprehensive quality score. It is a theoretical quality grade based on alcohol content;

[0099] The wine-picking decision unit dynamically adjusts the wine-picking segment points based on the time series of the comprehensive quality score, and outputs the corrected segment time points and the quality level of each segment.

[0100] The iterative update unit updates the weighting coefficients and ideal concentration ranges of flavor parameters based on the segmented deviation data.

[0101] In some embodiments, the iterative update unit described above includes:

[0102] The feedback collection subunit obtains the actual decision data of the wine picker on the segmentation point of this batch of wine picking, including the actual segmentation time point t1;

[0103] Deviation calculation sub-unit, calculates segmented deviation:

[0104] (7)

[0105] Where t2 represents the segmented time points of the model's predicted output;

[0106] The weight update sub-unit updates the weight coefficients based on the segmentation deviation and the deviation of each flavor parameter at the deviation time point:

[0107] (8)

[0108] in, It is the weight coefficient of the i-th flavor parameter after this iteration update. It is the weighting coefficient of the i-th flavor parameter used in the current batch. This is the learning rate, with a value ranging from 0.05 to 0.1. It is the time when the i-th flavor parameter deviates. The degree of deviation It indicates the time and location at which the segmented deviation occurs.

[0109] Secondly, embodiments of this application provide a real-time analysis system for wine components, which is constructed based on the construction method of the real-time analysis system for wine components in any one of the embodiments of the first aspect.

[0110] Thirdly, a dynamic wine extraction method for real-time analysis of wine components includes:

[0111] T10. Real-time spectral acquisition of the entire process of base liquor extraction;

[0112] T20. The collected data is sent to the second aspect of the real-time analysis system for wine components to obtain the wine extraction results.

[0113] In a specific example, the following steps can be taken:

[0114] Step 1: Confirmation of Base Liquor Grading Indicators

[0115] Daily testing data of base liquor from the distillery is collected, and process and quality experts are organized to conduct statistical analysis on the data to clarify the core indicators for base liquor grading, including seven indicators: alcohol content, total acid, total esters, acetic acid, lactic acid, ethyl acetate, and ethyl lactate. The grading thresholds for each indicator are determined to be compatible with the distillery's current base liquor quality and flavor standards, serving as the core benchmark for subsequent spectral modeling, distillation judgment, and quality grading.

[0116] Key parameters: The grading thresholds are implemented according to the winery's current base wine grading and flavor standards, without any additional hardware or environmental parameter controls.

[0117] Step 2: Preparation and debugging of near-infrared spectroscopy acquisition device

[0118] A near-infrared spectroscopy acquisition device was designed and fabricated. This device is adapted to the on-site environment of the distillery's distillation production line and can realize online real-time acquisition and spectral detection of base liquor samples. It ensures that the temperature of the liquor sample is consistent with the actual liquor receiving temperature. After completing the device debugging, the accuracy and stability of spectral acquisition are ensured, meeting the needs of continuous liquor collection and detection.

[0119] Key parameters: ① Sample temperature control: Water bath insulation is used, and the temperature is consistent with the actual wine receiving temperature, with a standard control of 40±2℃; ② Sample volume per test: Not less than 100ml; ③ Device debugging parameters: Spectral acquisition stability error ≤±0.02Abs, detection response time ≤1s; ④ Device protection: Protection level IP65, suitable for high temperature (≤80℃), high humidity (≤90%RH), and steam environments, and the explosion-proof level meets workshop safety standards.

[0120] Step 3: Collection, screening, and standard testing of representative wine samples

[0121] Systematic sampling of liquor was conducted at various receiving points in the distillery, covering different production shifts, batches, and fermentation pits. Sampling was carried out at each receiving point in three stages: the beginning, middle, and end of the distillation process, to ensure the comprehensiveness and representativeness of the samples. After collection, the KS algorithm was used to screen all sample spectra and extract representative samples. The standard testing methods of the liquor industry were used to accurately test seven core indicators of the screened representative samples to obtain standard physicochemical data for each sample, which served as a reference benchmark for the construction of the near-infrared spectral model.

[0122] Key parameters: ① Sampling distribution: Two samples are taken consecutively from the head of each distillation point, five samples are taken from the middle of the distillation (sampling interval is set according to distillation time, usually 1 minute / sample), and three samples are taken from the tail of the distillation, covering all distillation points in the same workshop; ② Total sample volume: No less than 300 samples are collected, and approximately 100 representative samples are selected after screening (accounting for 33% of the total sample volume); ③ Sample preservation: Sealed storage, temperature controlled at 20~25℃, avoiding direct sunlight, storage time not exceeding 72 hours, spectral acquisition is completed within 24 hours after sampling, and physicochemical index testing is completed within 72 hours; ④ Testing environment: Laboratory temperature 40±1℃, humidity 50±5%; ⑤ Testing accuracy: Alcohol content detection error ≤±0.5%vol, the detection errors of the other six indicators meet the current standards of the liquor industry, each sample is tested in parallel 3 times, and the average value is taken as the final standard data, with a parallel error ≤±5%.

[0123] Step 4: Construction and optimization of near-infrared spectroscopy detection model

[0124] The near-infrared spectral data of the screened samples were correlated and matched with the standard physicochemical data obtained in step 3. Chemometric algorithms (least squares PLS, principal component analysis PCA, support vector machine SVM) were used to construct a multi-component quantitative correction model and a qualitative discrimination model. The model was validated by dividing the modeling set and the validation set. A model iteration mechanism was introduced to continuously adjust the algorithm parameters, improve the model detection accuracy and adaptability, and ensure that the model can accurately detect the seven core indicators of base wine and complete the quality grading.

[0125] Key parameters: ① Ratio of modeling set to validation set: Modeling set accounts for 90%, validation set accounts for 10%; ② Model accuracy: Simultaneous detection accuracy of seven core indicators ≥90%, model determination coefficient R² ≥0.93, quality grading accuracy ≥95%, abnormal wine identification accuracy ≥98%, validation set deviation ≤±3%; ③ Algorithm parameters: PLS algorithm iteration count ≤300 times, PCA principal components are selected from the top 5~8 (cumulative contribution rate ≥95%), SVM algorithm penalty coefficient C=1~10, RBF kernel is selected as the kernel function; ④ Iteration cycle: The model is updated once per quarter based on production data and process adjustments to ensure model stability, and the detection deviation of base wine from different distillers and different batches is ≤±1%.

[0126] Step 5: Construction and debugging of the intelligent wine extraction system

[0127] High-precision near-infrared spectroscopy detection equipment (preferably grating equipment), data storage and computing servers, pipeline diversion devices (including intelligent valves and multi-channel storage tanks), and intelligent control systems are uniformly installed on all distillation production lines in the winery to build a fully intelligent distillation system. This system integrates modules such as spectral detection, multi-component analysis, quality grading, and intelligent diversion. After system debugging, the system ensures that the equipment is interconnected and operates stably, achieving a distillation process that requires no manual intervention.

[0128] Key parameters: ① Equipment parameters: Spectroscopic detection equipment accuracy ≥ 0.005 Abs, server data processing speed ≥ 10 records / second, intelligent valve response time ≤ 0.5s; ② Equipment consistency: Accuracy deviation of all spectral detection equipment ≤ ±0.01 Abs, intelligent valve response time uniformly ≤ 0.5s; ③ System compatibility: All equipment is fully compatible with the data management platform and intelligent control system, data transmission delay ≤ 0.5s; ④ Debugging standards: The system runs continuously for 100 hours without failure, abnormal alarm response time ≤ 10s, alarm accuracy ≥ 95%, accuracy of different grades of wine diversion ≥ 95%, avoiding mixed storage of premium and general wines.

[0129] Step 6: Execution of the entire intelligent wine extraction process

[0130] The intelligent liquor extraction system is activated, and near-infrared spectroscopy detection equipment collects spectra in real time throughout the entire process of base liquor extraction. The collected data is transmitted to the server, and through the optimization model constructed in step 4, the system automatically completes the detection and quality grading of seven core indicators of the base liquor. Based on the grading results, the intelligent control system automatically controls the intelligent valves in the pipeline to divert base liquor of different qualities to the corresponding storage tanks, realizing the full automation of base liquor extraction, grading, and diversion without human intervention, achieving the goal of "unmanned" quantity and quality extraction.

[0131] Key parameters: ① Detection response time: ≤1s, real-time detection, real-time decision-making, and real-time diversion throughout the process; ② Automation accuracy: quality judgment accuracy ≥95%, diversion accuracy 99%; ③ Operational stability: equipment failure rate ≤1% / month, system misjudgment rate ≤5%; ④ Quality target: stable improvement of base wine quality rate ≥1%, achieving the "superior grade, superior storage" target.

[0132] Step 7: Base Wine Data Management and Closed-Loop Optimization

[0133] Through the data management platform, key information such as spectral data, component data, quality grading results, and distillation time of each batch of base liquor is comprehensively recorded, creating a "digital passport" for each batch of base liquor. The base liquor data is connected to the distillery's digital blending system, providing traceable and quantifiable data support for digital blending. Each quarter, based on the base liquor data and blending feedback, the distillation model parameters and system operating parameters are optimized to form a closed loop of "production-data-optimization," continuously improving the quality and efficiency of distillation.

[0134] Key parameters: ① Data management: 100% completeness of digital passport information for base liquor, with data storage time ≥ 5 years; ② Data integration: 100% accuracy of data transmission with the digital blending system, with a latency ≤ 0.5s; ③ Closed-loop optimization: Regular detection and optimization when brewing process changes, ensuring method adaptability and practicality.

[0135] The real-time wine component analysis system, construction method, and dynamic wine extraction method proposed in this application have the following advantages:

[0136] I. Significantly Improved Performance: Relying on near-infrared spectroscopy multi-component detection technology, the accuracy of simultaneous detection of 7 core indicators is ≥90%, the grading accuracy is ≥95%, and the detection response time is ≤1s, greatly reducing the error of manual judgment (from 15%-20% to below 5%); the whole process is automated and closed-loop operation, the diversion accuracy is ≥99%, the equipment failure rate is ≤1% / month, and the quality rate of base wine is improved by ≥1%.

[0137] Second, production costs are significantly reduced: there is no need to rely on senior wine pickers, saving more than 70% of talent training costs, greatly reducing on-site operators and reducing labor costs; equipment maintenance is simple, base wine waste is reduced, and overall operating costs are reduced by more than 10% compared with the traditional model.

[0138] III. Convenient and efficient production: After startup, it operates fully automatically, and operators only need to master basic maintenance skills; it supports remote control and automatic data analysis, adapts to large-scale industrial production, and does not require large-scale modification of existing processes.

[0139] IV. Convenient processing and installation: The supporting equipment has a reasonable structure and mature processing technology, and can be mass-produced; the equipment protection level is IP65, which is suitable for the high temperature and high humidity steam environment of wineries. The installation does not require large-scale modification of the existing production line, the commissioning cycle is short, the compatibility is strong, and it can be quickly put into use.

[0140] V. Data and Intelligent Value-Added Advantages: Constructing a "digital passport" for base liquor, with 100% data integrity and a storage time of ≥5 years, solving the problem of the black box of traditional liquor extraction data and achieving full traceability of the liquor extraction process; seamlessly connecting base liquor data with the digital blending system, with 100% transmission accuracy and a latency of ≤0.5s, helping to upgrade the blending process with intelligence; introducing a quarterly closed-loop optimization mechanism, with the model detection accuracy improved by ≥2% after iteration, ensuring a stable increase of ≥1% in the rate of high-quality liquor, promoting the upgrading of baijiu brewing from intelligent production to digital management and refined optimization, and strengthening the core competitiveness of the industry.

[0141] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A construction method of a wine liquid component real-time analysis system, characterized by, include: S10. Construct a base wine grading index, wherein each index corresponds to a flavor parameter set, wherein the flavor parameter set includes the grading threshold range of at least one of alcohol content, total acid, total esters, acetic acid, lactic acid, ethyl acetate, and ethyl lactate. S20. Debug the near-infrared spectral acquisition device and adjust the various parameters of the near-infrared spectral acquisition device; S30. Obtain representative samples, and based on standard detection methods, detect various indicators of the representative samples to obtain standard physicochemical data of the representative samples, and use the standard physicochemical data as a reference standard for application in infrared spectroscopy acquisition devices. S40. The adjusted near-infrared spectrum is used to detect various indicators of the representative sample, and the detection results are correlated and matched with the standard physicochemical data. The correlated and matched data is sent to the iterative model for iteration to obtain the optimized parameters of the near-infrared spectral acquisition device.

2. The method of claim 1, wherein the wine component real-time analysis system is constructed by: The detection of various indicators of representative samples using adjusted near-infrared spectroscopy includes the following steps: S401, ensuring that the representative sample to be tested each time is not less than 80-120ml; S402. Keep the representative sample warm so that the temperature of the representative sample is consistent with the actual wine receiving temperature. S403. Perform tests on representative samples for various indicators.

3. The method of claim 1, wherein the wine component real-time analysis system is constructed by: Obtaining representative samples includes the following steps: S301. Collect samples from each receiving point in the winery, covering different production shifts, different batches, and different cellars. Samples are taken from each receiving point in three stages: the beginning, middle, and end of the wine. S302. Screen all sample spectra and extract representative samples.

4. The method for constructing the real-time analysis system for wine components as described in claim 3, characterized in that, The number of samples collected from each wine inlet in the winery shall not be less than 300, and the number of samples after spectral screening shall not be less than 100.

5. The method for constructing the real-time analysis system for wine components as described in claim 3, characterized in that, After collecting samples from each inlet within the winery, and before screening all sample spectra, the following steps are also included: The samples should be sealed and stored at a temperature of 20-25℃.

6. The method for constructing the real-time analysis system for wine components as described in claim 1, characterized in that, Iterative models include: The flavor quality calculation unit calculates the flavor quality index and the deviation of each flavor parameter based on the flavor parameter set. （1） Where F(t) is the flavor quality index, Xi is the i-th flavor parameter, and d Xi Let be the deviation of the i-th flavor parameter at time t, and wi be the weighting coefficient of the i-th flavor parameter. ; （2） （3） （4） in, and These represent the lower and upper limits of the ideal concentration of parameter X at alcohol content A, respectively. The correction unit calculates the correction magnitude based on the difference between the flavor quality index and the historical average flavor quality index, and outputs a comprehensive quality score: （5） in, The correction magnitude is α, and the correction coefficient is α. This represents the historical average flavor quality index at alcohol content A. （6） in, It is a comprehensive quality score. It is a theoretical quality grade based on alcohol content; The wine-picking decision unit dynamically adjusts the wine-picking segment points based on the time series of the comprehensive quality score, and outputs the corrected segment time points and the quality level of each segment. The iterative update unit updates the weighting coefficients and ideal concentration ranges of flavor parameters based on the segmented deviation data.

7. The method for constructing the real-time analysis system for wine components as described in claim 6, characterized in that, The iterative update unit includes: The feedback collection subunit obtains the actual decision data of the wine picker on the segmentation point of this batch of wine picking, including the actual segmentation time point t1; Deviation calculation sub-unit, calculates segmented deviation: （7） Where t2 represents the segmented time points of the model's predicted output; The weight update sub-unit updates the weight coefficients based on the segmentation deviation and the deviation of each flavor parameter at the deviation time point: （8） in, It is the weight coefficient of the i-th flavor parameter after this iteration update. It is the weighting coefficient of the i-th flavor parameter used in the current batch. This is the learning rate, with a value ranging from 0.05 to 0.

1. It is the time when the i-th flavor parameter deviates. The degree of deviation It indicates the time and location at which the segmented deviation occurs.

8. A real-time analysis system for wine components, characterized in that, It is constructed based on the construction method of the real-time analysis system of wine components according to any one of claims 1-7.

9. A dynamic wine extraction method for real-time analysis of wine components, characterized in that, include: T10. Real-time spectral acquisition of the entire process of base liquor extraction; T20. The collected data is sent to the real-time analysis system of wine components as described in claim 8 to obtain the wine extraction results.

Images