Wort oxygenation speed intelligent control method and system for beer production

Abstract

The invention discloses an intelligent wort oxygenation speed control method and system for beer production, belongs to the technical field of beer production, and dynamically predicts a theoretical dissolved oxygen concentration value based on a random forest regression model constructed based on multivariable real-time parameters and historical fermentation data. The deviation score is calculated by fusing the drift value of the dissolved oxygen sensor and the abnormal value of the instantaneous oxygen feeding amount, so that the intelligent evaluation of the severity of the system deviation is realized; when the deviation score exceeds a threshold value, reference flow meter data is automatically stopped, and control logic is driven by a model output value; meanwhile, abnormal fluctuation of the real-time dissolved oxygen concentration in a target range is analyzed, the opening degree of an oxygen valve and the oxygen supply pressure are intelligently adjusted, and it is ensured that the oxygenation process is stable and reliable; according to the technology, the problem of quality fluctuation caused by sensor error accumulation is effectively solved, and the beer flavor consistency and the intelligent level of system operation are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of beer production, in particular to a wort oxygenation speed intelligent control method and system for beer production. BACKGROUND

[0002] The wort oxygenation speed intelligent control for beer production refers to intelligently adjusting the oxygen injection speed in the wort during the beer brewing process to achieve the control technology of the best fermentation effect. The wort oxygenation is a key step in brewing, which directly affects the yeast activity and fermentation quality. Through intelligent sensors and algorithm systems, the oxygen injection amount and speed are automatically adjusted according to the real-time monitoring of the temperature, concentration, flow rate and other parameters of the wort, so as to improve the brewing efficiency and ensure the stability of beer flavor and product consistency.

[0003] The prior art has the following shortcomings: During the wort oxygenation process of beer, if the oxygen mass flowmeter is not calibrated for a long time, zero drift or sensitivity degradation may occur, which makes the oxygen flow data obtained by the PLC low, and the dissolved oxygen is judged as insufficient. At this time, the system will continuously increase the oxygen supply, resulting in over-saturation of oxygen in the wort (such as exceeding 15 ppm). This excessive oxygenation can cause yeast stress or even mutation, which in turn causes fermentation abnormalities or failure, and generates a large amount of by-products such as high-alcohol and aldehyde, which seriously damages the flavor of beer. In addition, the enhanced oxidation reaction also significantly shortens the shelf life of the product. Since the error accumulation of the flowmeter is gradual, it is difficult to detect in the short term, and it is often indirectly identified after the product quality fluctuates, and the traceability and correction have high technical difficulty. SUMMARY

[0004] The purpose of the present application is to provide a wort oxygenation speed intelligent control method and system for beer production to solve the problems in the background art.

[0005] In order to achieve the above purpose, the present application provides the following technical scheme: a wort oxygenation speed intelligent control method for beer production, comprising: obtaining real-time parameters of wort in the oxygenation process; inputting the real-time parameters into a multivariate nonlinear regression model, the model being constructed based on historical wort fermentation data, adopting a random forest regression algorithm, and predicting the theoretical dissolved oxygen concentration value of the wort under the current working condition; comprehensively calculating the dissolved oxygen concentration deviation score value by obtaining the dissolved oxygen sensor drift value and the oxygen instantaneous addition amount abnormal value to evaluate the deviation severity of the theoretical dissolved oxygen concentration value output by the model and the actual dissolved oxygen sensor feedback value; When the dissolved oxygen concentration deviation score value exceeds a preset threshold value, stopping referring to oxygen mass flow meter data, and the dissolved oxygen estimate value output by the prediction model is used as a control reference value; An abnormal fluctuation of the real-time dissolved oxygen concentration within a target set value range is analyzed, and the opening of the oxygen electric regulating valve and the oxygen supply pressure are automatically adjusted according to the analysis result.

[0006] Preferably, the real-time parameters include wort flow rate, temperature, oxygen supply pressure, oxygen valve opening and dissolved oxygen concentration before and after oxygenation.

[0007] Preferably, the multivariate nonlinear regression model is a random forest regression model, which includes: combining the collected real-time parameters and historical wort fermentation batch data to form a training data set; dividing the data into a training set and a test set to construct a random forest model; setting model parameters, including the number of decision trees, the maximum depth and the minimum sample division number; minimizing the mean square error between the predicted value and the actual dissolved oxygen value as the target for model training, and outputting the theoretical dissolved oxygen concentration value.

[0008] Preferably, the method for obtaining the dissolved oxygen sensor drift value is: Set the window length W, which represents the time length used for fitting each time, and set the sliding step S, which represents the time interval of each window forward movement; for the current sliding window, extract the data sequence: ; wherein: represents the time stamp; represents the dissolved oxygen concentration value at time ; n represents the number of sampling points in the window, and the data points in the window are subjected to least square linear fitting to obtain a fitting straight line: y = βt + α; wherein β is the regression slope, representing the change rate of the dissolved oxygen concentration per unit time; α is the intercept, representing the estimated value of the starting point; define the dissolved oxygen sensor drift value DV, and the expression is: .

[0009] Preferably, the method for obtaining the oxygen instantaneous addition amount abnormal value is: define the time window size M, for each time t, construct a window containing the current value and the previous M-1 values: ; calculate the median and the absolute deviation, and the median The calculation expression of the median absolute deviation is: ; calculate the median absolute deviation set , the expression is: ; for the current time point t, define the normalized deviation as: ; wherein: k is a conventional value of 1.4826; to prevent the denominator from being zero, is the MAD abnormal score; set an abnormal threshold T; if T, the current oxygen instantaneous addition amount is considered as an abnormal value; the value of this time is taken as the oxygen instantaneous addition amount abnormal value.

[0010] Preferably, the dissolved oxygen sensor drift value and the oxygen instantaneous addition amount abnormal value are converted into a comprehensive feature vector, the comprehensive feature vector is taken as the input of a machine learning model, the machine learning model takes the dissolved oxygen concentration deviation score value label as the prediction target, the training target is to minimize the sum of prediction errors of all dissolved oxygen concentration deviation score value labels, the machine learning model is trained until the sum of prediction errors converges, and the model training is stopped, and the dissolved oxygen concentration deviation score value is determined according to the model output result, wherein the machine learning model is a weighted summation model.

[0011] Preferably, the obtained dissolved oxygen concentration deviation score value is compared with a preset threshold value, if the dissolved oxygen concentration deviation score value is greater than or equal to the preset threshold value, it indicates that the deviation severity of the theoretical dissolved oxygen concentration value output by the model and the actual dissolved oxygen sensor feedback value is high, at this time a warning signal is generated; if the dissolved oxygen concentration deviation score value is less than the preset threshold value, it indicates that the deviation severity of the theoretical dissolved oxygen concentration value output by the model and the actual dissolved oxygen sensor feedback value is low, at this time no warning signal is generated.

[0012] Preferably, the oxygen electric regulating valve opening and the oxygen supply pressure are dynamically adjusted, and the specific method comprises the following steps: ; wherein: , wherein, is the actual dissolved oxygen concentration at t, is the actual dissolved oxygen concentration at t, is the actual dissolved oxygen concentration at t, λ is a fluctuation sensitivity coefficient, is the defined abnormal fluctuation factor, is the target set value, is the standard deviation of the actual dissolved oxygen concentration; The oxygen electric regulating valve opening ΔV is obtained by weighting and summing the dissolved oxygen concentration deviation score value and the abnormal fluctuation factor after de-dimensioning and normalizing processing; if ΔV>0: increase the opening, increase the oxygen supply; if ΔV<0: reduce the opening, reduce the oxygen supply; adjust the oxygen supply pressure ΔP: ; is the oxygen supply pressure adjustment value, γ is a pressure adjustment coefficient, is the dissolved oxygen concentration deviation score value.

[0013] The application also provides a wort oxygenation speed intelligent control system for beer production, which comprises a data acquisition module, a dissolved oxygen concentration prediction module, a deviation evaluation module, an intelligent control switching module and a dynamic control module.​ Data acquisition module: obtain real-time parameters of wort during oxygenation process; Dissolved oxygen concentration prediction module: input the real-time parameters into a multivariate nonlinear regression model, which is constructed based on historical wort fermentation data and uses a random forest regression algorithm, to predict the theoretical dissolved oxygen concentration value of wort under the current working condition; Deviation evaluation module: generate a dissolved oxygen concentration deviation score value by comprehensively calculating the dissolved oxygen sensor drift value and the oxygen instantaneous addition amount abnormal value, which is used to evaluate the severity of the deviation between the theoretical dissolved oxygen concentration value output by the model and the actual dissolved oxygen sensor feedback value; Intelligent control switching module: when the dissolved oxygen concentration deviation score value exceeds a preset threshold, stop referring to the oxygen mass flow meter data, and use the dissolved oxygen estimate value output by the prediction model as the control reference value; Dynamic control module: analyze abnormal fluctuations of real-time dissolved oxygen concentration within the target set value range, and automatically adjust the opening of the oxygen electric regulating valve and the oxygen supply pressure according to the analysis results.

[0014] In the above technical solution, the present application provides technical effects and advantages: 1. The present application predicts the theoretical dissolved oxygen concentration value by constructing a multivariate nonlinear regression model, and combines the dissolved oxygen sensor drift value and the oxygen instantaneous addition amount abnormal value for comprehensive evaluation, thereby achieving precise control of the wort oxygenation process. This method can effectively identify the zero drift and sensitivity degradation of the oxygen mass flow meter, dynamically switch the control reference value (model estimate value replaces the faulty sensor data), and intelligently adjust the opening of the oxygen valve and the supply pressure, thereby fundamentally avoiding the problems of yeast stress, abnormal fermentation and flavor imbalance caused by over-oxygenation, and significantly improving the beer quality stability and shelf life.

[0015] 2. The present application combines machine learning prediction models with real-time anomaly detection mechanisms, and realizes early warning and automatic fault tolerance control through the dissolved oxygen concentration deviation score value. Compared with traditional single-sensor dependent systems, this system has stronger anti-interference ability and self-adaptive characteristics, can intervene in time at the early stage of sensor performance degradation, avoid batch quality accidents caused by progressive error accumulation, reduce the frequency of manual calibration and maintenance costs, and provides reliable technical support for the intelligent upgrading of beer production. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments or prior art, the following will briefly introduce the drawings needed in the embodiments. Obviously, the drawings described below are only some embodiments described in the present application, and other drawings can also be obtained by those skilled in the art based on these drawings.

[0017] Figure 1 Mind map for the method of the present application.

[0018] Figure 2 Mind map for the system module of the present application. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0020] Embodiment 1, please refer to Figure 1 The method for intelligent control of wort oxygenation rate in beer production described in this embodiment comprises the following steps: Obtaining real-time parameters of wort in the oxygenation process; Inputting the real-time parameters into a multivariate nonlinear regression model, the model being constructed based on historical wort fermentation data, adopting a random forest regression algorithm, and predicting a theoretical dissolved oxygen concentration value of the wort under the current working condition; Generating a dissolved oxygen concentration deviation score value through comprehensive calculation of the dissolved oxygen sensor drift value and the oxygen instantaneous addition amount abnormal value, for evaluating the severity of the deviation between the theoretical dissolved oxygen concentration value output by the model and the actual dissolved oxygen sensor feedback value; When the dissolved oxygen concentration deviation score value exceeds a preset threshold value, stopping referring to the oxygen mass flowmeter data, and taking the dissolved oxygen estimate value output by the prediction model as a control reference value; Analyzing abnormal fluctuations of the real-time dissolved oxygen concentration within the target set value range, and automatically adjusting the opening degree of the oxygen electric regulating valve and the oxygen supply pressure according to the analysis result.

[0021] In the beer wort oxygenation control system, accurate acquisition of key process parameters is the basis of intelligent control. This step realizes online real-time acquisition and digital processing of core variables in the wort oxygenation stage by integrating various industrial-grade sensors and data acquisition equipment, specifically including the following aspects: An electromagnetic flowmeter or a mass flowmeter is installed in the wort conveying pipeline, usually at the front section of the oxygenation system; the flowmeter transmits flow rate data in real time to the PLC through 4~20mA analog signals or Modbus protocol; the flow rate data is used to determine the wort flux, decide the oxygen injection amount and mixing time, and is an important variable for subsequent modeling.

[0022] Pt100 temperature sensor is arranged on the wort circulation pipeline, and the measuring point is arranged at the front end before oxygenation and the discharge end of the mixer; the wort temperature is collected in real time and input into the PLC for dynamic correction of the oxygen dissolving capacity (the higher the temperature, the lower the oxygen dissolving capacity); the temperature data are also used to judge whether the cooling system is normally operated.

[0023] A pressure transmitter (for example, 0-6 bar range, 0.5 level precision) is installed on the oxygen supply pipeline; the transmitter is used to monitor the oxygen inlet pressure, so as to ensure that the mixer is in stable working condition; abnormal pressure fluctuation will trigger the interlock protection of the oxygen supply system as an alarm signal.

[0024] The oxygen regulating valve used is an electric regulating valve with feedback signal (with position transmitter); the opening percentage of the valve is fed back to the PLC through analog quantity as a direct control quantity of the oxygen supply amount; the data are associated with the oxygen flow estimation model and used to inversely deduce the actual oxygen supply rate.

[0025] Dissolved oxygen online sensors (using polarographic method or optical fluorescence method) are respectively installed before and after the mixer; the front sensor is used to monitor the initial oxygen content of the wort, and the rear sensor reflects the actual dissolved oxygen increment; the dissolved oxygen concentration is a core feedback signal of the closed-loop regulation of the PLC, and determines whether the opening of the oxygen valve is increased or decreased; the value is also used to compare with the prediction model to detect whether the sensor or flowmeter is inaccurate.

[0026] All parameters are integrated into the PLC master station through industrial Ethernet or Profibus protocol; the PLC periodically refreshes data every 1-2 seconds, writes into the cache and executes the control logic; the data are synchronously uploaded to the upper computer (SCADA system) for curve monitoring and historical tracing.

[0027] Through the specific implementation of the above steps, the control system can capture multiple dimensional variables affecting the dynamic change of dissolved oxygen at high frequency and high precision, and provide a comprehensive and reliable data basis for subsequent model prediction, deviation correction and closed-loop regulation. The highly integrated data acquisition scheme significantly improves the intelligent level of oxygenation control and the process stability.

[0028] The present application constructs and deploys a multivariate nonlinear regression model, uses historical wort fermentation data as the basis, adopts a random forest regression algorithm to process real-time parameters in the oxygenation process, predicts the theoretical dissolved oxygen concentration value that the wort should reach under the current working condition, and provides a decision basis for subsequent intelligent control. The specific technical process is as follows: The following multiple dimensional real-time parameters are collected and input as model characteristic variables (input vector): Wort flow rate (unit: L / h), wort temperature (unit: ℃), oxygen line pressure (unit: bar), oxygen valve opening (unit: %), dissolved oxygen concentration before oxygenation (unit: ppm), equipment operating time (as an implicit indicator of flow meter or sensor degradation), and fermenter model or batch number.

[0029] Using data from multiple historical fermentation batches, the above-mentioned real-time parameters and dissolved oxygen concentration under stable conditions after actual oxygenation were collected as target values ​​(labels); data cleaning and standardization were performed, including missing value imputation, outlier removal, and unit unification; the data were divided into training sets (e.g., 70%) and test sets (e.g., 30%) to evaluate the model's generalization ability.

[0030] Build a random forest regression model in a Python environment (such as the Scikit-learn library) or an industrial modeling tool (such as MATLAB or LabVIEW); set hyperparameters such as the number of trees in the forest (e.g., 100 trees), the maximum tree depth, and the minimum number of split samples; random forests model nonlinear relationships by integrating multiple decision trees and are suitable for the volatility and redundancy characteristics of industrial data.

[0031] The training set is input into the model for training, with the goal of minimizing the mean squared error (MSE) between the predicted and actual dissolved oxygen values. The model is then validated using a test set, with evaluation metrics including the coefficient of determination and the mean absolute error (MAE). If the model performance is insufficient, the parameters are further optimized through cross-validation or grid search.

[0032] The trained model is deployed on a PLC system or SCADA platform via an edge computing platform or industrial PC; real-time parameters are automatically received at regular sampling intervals (e.g., 5 seconds), and a prediction is performed; the model outputs the theoretical dissolved oxygen concentration value (unit: ppm) under the current operating conditions as a dynamic benchmark value.

[0033] This model leverages the nonlinear modeling capabilities and noise resistance of random forests to effectively capture the relationships between complex variables during wort aeration, enabling accurate prediction of dissolved oxygen concentration and enhancing the ability to detect abnormal operating conditions. Compared to traditional empirical control or single-point sensor control methods, it exhibits higher stability and robustness, making it suitable for the real-time intelligent adjustment needs of large-scale beer production environments.

[0034] A dissolved oxygen concentration deviation score is generated by comprehensively calculating the dissolved oxygen sensor drift value and the abnormal value of the instantaneous oxygen dosage. This score is used to assess the severity of the deviation between the theoretical dissolved oxygen concentration value output by the model and the actual dissolved oxygen sensor feedback value. Specifically, it includes: The method for obtaining the dissolved oxygen sensor drift value is as follows: Set the window length W, which represents the length of time (in minutes or number of sampling points) used for each fitting step, such as W=10 minutes. Set the sliding step S, which represents the time interval between each window movement forward. It is recommended to set S=1~2 minutes to ensure smooth and continuous operation.

[0035] For the current sliding window, extract the data sequence: ;in: Represents a timestamp (it is recommended to use a floating-point number converted from minutes or seconds); Indicates time The dissolved oxygen concentration value (unit: ppm); n represents the number of sampling points within the window, such as once per minute, n=W.

[0036] A least-squares linear fit is performed on the data points within the window, and the fitted line is: y = βt + α; where β is the regression slope, representing the rate of change of dissolved oxygen concentration per unit time (unit: ppm / min); α is the intercept, representing the estimated value of the starting point; the least-squares calculation expression is: , Define the dissolved oxygen sensor drift value DV as follows: When the DV exceeds the preset threshold within the window, it is determined that the sensor may be experiencing slow drift or abnormal response.

[0037] The method for obtaining abnormal values ​​of instantaneous oxygen dosage is as follows: Define the time window size M, for example, M = 15 minutes or 15 sampling points; for each time t, construct a window containing the current value and the previous M-1 values: ; Calculate the median and absolute deviation, median The calculation expression is: ; Calculate the set of absolute deviations of the median The expression is: For the current time point t, the standardized deviation is defined as: Where: k is a commonly used value of 1.4826 (MAD can be used as a standard deviation estimator when the data is close to a normal distribution); To prevent the denominator from being zero, it is set to extremely small; Assign an anomaly score to MAD; set an anomaly threshold T (e.g., T=3 or T=3.5); if If the value is greater than T, then the current instantaneous oxygen dosage is considered an anomaly; [the value will be displayed here]. The value is used as an abnormal value for the instantaneous oxygen dosage.

[0038] The dissolved oxygen sensor drift values ​​and abnormal values ​​of instantaneous oxygen dosage are converted into comprehensive feature vectors. These comprehensive feature vectors are used as input to a machine learning model. The machine learning model uses the predicted dissolved oxygen concentration deviation score label for each set of comprehensive feature vectors as the prediction objective and minimizes the sum of prediction errors for all dissolved oxygen concentration deviation score labels as the training objective. The machine learning model is trained until the sum of prediction errors converges, at which point the model training stops. The dissolved oxygen concentration deviation score is determined based on the model output. The machine learning model is a weighted summation model.

[0039] The obtained dissolved oxygen concentration deviation score is compared with a preset threshold. If the score is greater than or equal to the threshold, it indicates a high degree of deviation between the theoretical dissolved oxygen concentration output by the model and the actual dissolved oxygen sensor feedback value, and an early warning signal is generated. If the score is less than the threshold, it indicates a low degree of deviation, and no early warning signal is generated. Here, δ is an empirically set preset threshold, such as 1.0 ppm.

[0040] When the dissolved oxygen concentration deviation score exceeds a preset threshold: the output data of the reference oxygen mass flow meter is stopped. The system determines that the current oxygen flow meter may have deviation distortion or range drift. To avoid interference with the automatic control results, the control system temporarily shields the flow meter data from affecting the oxygen supply regulation.

[0041] The dissolved oxygen estimate output by the predictive model is used as an alternative control reference value; the control system uses the dissolved oxygen concentration estimate calculated by the trained multivariate regression model as the benchmark target value for the current oxygenation regulation.

[0042] The system compares the estimated dissolved oxygen value with the target dissolved oxygen range, and based on the error value, continues to drive closed-loop controllers such as PID controllers to adjust execution variables such as oxygen valve opening and compressed gas volume to maintain the dissolved oxygen in the wort within the set range (e.g., 8–10 ppm).

[0043] The control system records this switchover as an abnormal operating event and sends a notification to the operation and maintenance system, prompting operators to calibrate, clean, or replace the oxygen mass flow meter.

[0044] This invention, under the premise that the dissolved oxygen concentration is within the target setting range in real time, further identifies its short-term abnormal fluctuation behavior, and combines it with the previously calculated dissolved oxygen concentration deviation score value to dynamically adjust the opening of the oxygen electric regulating valve and the oxygen supply pressure, thereby achieving more precise oxygen supply control.

[0045] The abnormal volatility factor is defined as: ;in: ,in, The actual dissolved oxygen concentration (ppm) at time t. for The actual dissolved oxygen concentration (ppm) at any given time, λ is the fluctuation sensitization coefficient (recommended value 1~2), to enhance the responsiveness to high-frequency fluctuations. To define the abnormal volatility factor, Set a target value (ppm), such as 9.0. This represents the standard deviation of the actual dissolved oxygen concentration.

[0046] Introducing the dissolved oxygen concentration deviation score calculated earlier As a criterion for judging the severity of the overall system deviation, combined with the current volatility factor Develop dynamic control strategies: Adjusting the opening degree ΔV of the oxygen electric regulating valve: After dimensionless normalization of the obtained dissolved oxygen concentration deviation score and abnormal fluctuation factor, the opening degree ΔV of the oxygen electric regulating valve is calculated by weighted summation; if ΔV>0: increase the opening degree to increase oxygen supply; if ΔV<0: decrease the opening degree to reduce oxygen supply.

[0047] Adjust the oxygen supply pressure ΔP: ; The oxygen supply pressure adjustment value (unit: bar) is given, and γ is the pressure adjustment coefficient (e.g., 0.05 to 0.2). The product form ensures that the pressure is adjusted only when fluctuations and systematic deviations exist simultaneously, preventing frequent fluctuations from affecting the stability of the gas supply system.

[0048] Example 2, please refer to Figure 2 As shown in this embodiment, an intelligent control system for wort oxygenation speed in beer production includes a data acquisition module, a dissolved oxygen concentration prediction module, a deviation assessment module, an intelligent control switching module, and a dynamic control module. Data acquisition module: Acquires real-time parameters of wort during the oxygenation process; Dissolved oxygen concentration prediction module: The real-time parameters are input into a multivariate nonlinear regression model. The model is constructed based on historical wort fermentation data and uses a random forest regression algorithm to predict the theoretical dissolved oxygen concentration of the wort under the current operating conditions. Deviation assessment module: By acquiring the dissolved oxygen sensor drift value and the abnormal value of instantaneous oxygen dosage, a comprehensive calculation is performed to generate a dissolved oxygen concentration deviation score, which is used to assess the severity of the deviation between the theoretical dissolved oxygen concentration value output by the model and the actual dissolved oxygen sensor feedback value. Intelligent control switching module: When the dissolved oxygen concentration deviation score exceeds the preset threshold, the reference oxygen mass flow meter data is stopped, and the dissolved oxygen estimate output by the prediction model is used as the control reference value; Dynamic control module: Analyzes abnormal fluctuations in real-time dissolved oxygen concentration within the target set value range, and automatically adjusts the opening of the oxygen electric regulating valve and the oxygen supply pressure based on the analysis results.

[0049] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0050] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0051] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0052] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for intelligent control of wort oxygenation rate in beer production, characterized in that: include: Obtain real-time parameters of the wort during the oxygenation process; The real-time parameters are input into a multivariate nonlinear regression model, which is built based on historical wort fermentation data and uses a random forest regression algorithm to predict the theoretical dissolved oxygen concentration of wort under the current operating conditions. A dissolved oxygen concentration deviation score is generated by comprehensively calculating the drift value of the dissolved oxygen sensor and the abnormal value of the instantaneous oxygen dosage. This score is used to evaluate the severity of the deviation between the theoretical dissolved oxygen concentration value output by the model and the actual dissolved oxygen sensor feedback value. When the dissolved oxygen concentration deviation score exceeds the preset threshold, the reference oxygen mass flow meter data is stopped, and the dissolved oxygen estimate output by the prediction model is used as the control reference value. The system analyzes abnormal fluctuations in real-time dissolved oxygen concentration within the target set range and automatically adjusts the opening of the oxygen electric regulating valve and the oxygen supply pressure based on the analysis results.

2. The intelligent control method for wort oxygenation speed in beer production according to claim 1, characterized in that: The real-time parameters include wort flow rate, temperature, oxygen supply pressure, oxygen valve opening, and dissolved oxygen concentration before and after oxygenation.

3. The intelligent control method for wort oxygenation rate in beer production according to claim 1, characterized in that: The multivariate nonlinear regression model is a random forest regression model, which includes: combining the collected real-time parameters with historical wort fermentation batch data to form a training dataset; dividing the data into training and testing sets to construct a random forest model; setting model parameters, including the number of decision trees, maximum depth, and minimum number of sample splits; training the model with the goal of minimizing the mean square error between the predicted value and the actual dissolved oxygen value, and outputting the theoretical dissolved oxygen concentration value.

4. The intelligent control method for wort oxygenation speed in beer production according to claim 3, characterized in that: The method for obtaining the dissolved oxygen sensor drift value is as follows: Set the window length W to represent the time length used for fitting each time, and set the sliding step S to represent the time interval for each window movement; for the current sliding window, extract the data sequence: ;in: Represents a timestamp; Indicates time The dissolved oxygen concentration value; n represents the number of sampling points within the window. A least-squares linear fit is performed on the data points within the window, and the fitted line is: y = βt + α; where β is the regression slope, representing the rate of change of dissolved oxygen concentration per unit time; α is the intercept, representing the estimated value of the starting point; the dissolved oxygen sensor drift value DV is defined as: .

5. The intelligent control method for wort oxygenation speed in beer production according to claim 4, characterized in that: The method for obtaining outliers in instantaneous oxygen dosage is as follows: Define a time window size M, and for each time t, construct a window containing the current value and the previous M-1 values: ; Calculate the median and absolute deviation, median The calculation expression is: ; Calculate the set of absolute deviations of the median The expression is: For the current time point t, the standardized deviation is defined as: Where: k is the commonly used value of 1.4826; To prevent the denominator from being zero, MAD anomaly score; Set an abnormal threshold T; if If the value is greater than T, then the current instantaneous oxygen dosage is considered an anomaly; [the value will be displayed here]. The value is used as an abnormal value for the instantaneous oxygen dosage.

6. The intelligent control method for wort oxygenation speed in beer production according to claim 5, characterized in that: The dissolved oxygen sensor drift values ​​and abnormal values ​​of instantaneous oxygen dosage are converted into comprehensive feature vectors. These comprehensive feature vectors are used as input to a machine learning model. The machine learning model uses the predicted dissolved oxygen concentration deviation score label for each set of comprehensive feature vectors as the prediction objective and minimizes the sum of prediction errors for all dissolved oxygen concentration deviation score labels as the training objective. The machine learning model is trained until the sum of prediction errors converges, at which point the model training stops. The dissolved oxygen concentration deviation score is determined based on the model output. The machine learning model is a weighted summation model.

7. The intelligent control method for wort oxygenation rate in beer production according to claim 6, characterized in that: The obtained dissolved oxygen concentration deviation score is compared with a preset threshold. If the dissolved oxygen concentration deviation score is greater than or equal to the preset threshold, it indicates that the deviation between the theoretical dissolved oxygen concentration value output by the model and the actual dissolved oxygen sensor feedback value is severe, and an early warning signal is generated. If the dissolved oxygen concentration deviation score is less than the preset threshold, it indicates that the deviation between the theoretical dissolved oxygen concentration value output by the model and the actual dissolved oxygen sensor feedback value is mild, and no early warning signal is generated.

8. The intelligent control method for wort oxygenation rate in beer production according to claim 7, characterized in that: Dynamically adjusting the opening of the oxygen electric regulating valve and the oxygen supply pressure specifically includes defining an abnormal fluctuation factor as follows: ;in: ,in, Let be the actual dissolved oxygen concentration at time t. for The actual dissolved oxygen concentration at time t, where λ is the fluctuation sensitization coefficient. To define the abnormal volatility factor, Set a value for the target. This represents the standard deviation of the actual dissolved oxygen concentration. Adjusting the oxygen electric regulating valve opening ΔV: After dimensionlessly normalizing the obtained dissolved oxygen concentration deviation score and abnormal fluctuation factor, the oxygen electric regulating valve opening ΔV is calculated by weighted summation; if ΔV>0: increase the opening to increase oxygen supply; if ΔV<0: decrease the opening to reduce oxygen supply; Adjusting the oxygen supply pressure ΔP: ; γ is the oxygen supply pressure regulation value, and γ is the pressure regulation coefficient. This is the score for the deviation in dissolved oxygen concentration.

9. An intelligent control system for wort oxygenation speed in beer production, used to implement the intelligent control method for wort oxygenation speed in beer production as described in any one of claims 1-8, characterized in that: It includes a data acquisition module, a dissolved oxygen concentration prediction module, a deviation assessment module, an intelligent control switching module, and a dynamic control module; Data acquisition module: Acquires real-time parameters of wort during the oxygenation process; Dissolved oxygen concentration prediction module: The real-time parameters are input into a multivariate nonlinear regression model. The model is constructed based on historical wort fermentation data and uses a random forest regression algorithm to predict the theoretical dissolved oxygen concentration of the wort under the current operating conditions. Deviation assessment module: By acquiring the dissolved oxygen sensor drift value and the abnormal value of instantaneous oxygen dosage, a comprehensive calculation is performed to generate a dissolved oxygen concentration deviation score, which is used to assess the severity of the deviation between the theoretical dissolved oxygen concentration value output by the model and the actual dissolved oxygen sensor feedback value. Intelligent control switching module: When the dissolved oxygen concentration deviation score exceeds the preset threshold, the reference oxygen mass flow meter data is stopped, and the dissolved oxygen estimate output by the prediction model is used as the control reference value; Dynamic control module: Analyzes abnormal fluctuations in real-time dissolved oxygen concentration within the target set value range, and automatically adjusts the opening of the oxygen electric regulating valve and the oxygen supply pressure based on the analysis results.

Images