Automatic liquor picking control method and system

Abstract

The invention discloses an automatic wine picking control method and system. The method comprises the steps that buoyancy data, temperature data and liquid level data of flowing wine flowing into a detection device are collected in real time; based on the buoyancy data and the liquid level data, calculating the real-time density of the flowing wine by combining preset calibration parameters, and obtaining an initial alcohol degree value according to a mapping relation between the density and the alcohol degree; performing temperature compensation on the initial alcohol value according to the temperature data to obtain a compensated alcohol value; based on the liquid level data, real-time volume flow is obtained according to a preset liquid level-flow calibration curve, and mass flow is calculated; at least one of the data collected in real time and the data obtained through calculation serves as a judgment object, the judgment object is compared with a preset section switching condition, and flowing wine in the detection device is guided into the container of the corresponding wine section. According to the method, the liquor in different liquor sections can be automatically picked out, the accumulated mass and the comprehensive alcohol content of the liquor in each section can be calculated, and the liquor picking efficiency and the liquor output stability and reliability are improved.

Description

Technical Field

[0001] This invention relates to the field of brewing technology, and in particular to an automatic wine extraction control method and system. Background Technology

[0002] In traditional baijiu brewing, the distillation process primarily relies on manual observation or the use of alcohol meters to read the alcohol content. Throughout the process, workers repeatedly use glass alcohol meters and thermometers to measure the current alcohol content of the baijiu, and also manually weigh each portion to determine its quality. This method requires workers to stay in the distillation area for extended periods, observing parameters such as the volume and temperature of the liquor, and then manually moving the liquid to designated areas. However, both manual observation and alcohol meter readings have inherent errors, resulting in low precision and accuracy in the distillation process. Furthermore, this method is highly dependent on the workers' experience, and the results can vary significantly between different workers. In addition, manual distillation is labor-intensive, time-consuming, and has low production efficiency.

[0003] The "Method for Extracting Distilled Baijiu of Mellow Type" disclosed in Chinese patent literature, publication number CN102559465B, includes the following steps: a) Loading the fermented lees into a still for distillation, with a cloth lining the mouth of the still forming a groove; the steam pressure is controlled at 0.02-0.03 MPa during distillation; b) After the distillation begins, extract a section of the liquor to remove aldehydes and mustard-like off-flavors; c) After extracting the first section, extract a second section of the liquor, observing the transition from foam to saliva foam. The process involves three steps: d) Extracting the first three segments of alcohol from the initial "saliva foam" (the foam is characterized by bubbles of varying sizes intermingled, persisting for 2-5 seconds); d) Extracting the third segment from the initial foam, breaking the foam when it stops; a broken foam is defined as bubbles less than 1mm in diameter that disappear immediately; e) Extracting the fourth segment when the alcohol content reaches zero; The flow rate in steps b, c, d, and e is 2-5 kg / min; the flow temperature is 25-35℃. This technique only standardizes the criteria for judging different segments of alcohol during extraction, but manual extraction and judgment are still required. It heavily relies on the experience of the workers, and the extraction results may vary significantly between different workers, leading to lower precision and accuracy. Summary of the Invention

[0004] The present invention aims to overcome the problems of existing technologies that require manual wine picking and judgment, which rely heavily on the wine picking experience of workers and result in significant differences in wine picking effects among different workers, leading to low precision and accuracy in wine picking. The invention provides an automatic wine picking control method and system.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: An automatic wine-collecting control method includes: Real-time data collection of buoyancy, temperature, and liquid level of the flowing wine into the detection device; Based on buoyancy data and liquid level data, combined with preset calibration parameters, the real-time density of the flowing wine is calculated, and the initial alcohol content value is obtained according to the mapping relationship between density and alcohol content; the initial alcohol content value is compensated for by temperature data to obtain the compensated alcohol content value. Based on the liquid level data, the real-time volumetric flow rate is obtained according to the preset liquid level-flow rate calibration curve, and the mass flow rate is calculated. At least one of the real-time collected data and the calculated data is used as the judgment object and compared with the preset transition conditions to guide the flowing wine in the detection device to the container of the corresponding wine segment.

[0006] This invention utilizes real-time monitoring by multiple sensors and corresponding measurement structure design, combined with pre-set judgment conditions for different liquor segments, to extract liquor from different segments and calculate the cumulative mass and overall alcohol content of each segment, thereby improving extraction efficiency and the stability and reliability of liquor output. For the initial alcohol content obtained through density mapping, compensation correction is performed using the liquor temperature to obtain a more accurate real-time alcohol content value. At the same time, the method for obtaining the real-time flow rate of the liquor is optimized, resulting in a more accurate flow rate result and further improving the accuracy of automatic liquor extraction.

[0007] Preferably, the calculation of the real-time density of the flowing wine based on buoyancy data and liquid level data, combined with preset calibration parameters, includes: The real-time density of the liquid is calculated by adding the volume error compensation value to the ratio of the buoyancy calibration value to the density of the calibration liquid, and using the buoyancy data as the numerator. The volume error compensation value is calculated by the difference between the liquid level data and the liquid level calibration value.

[0008] Preferably, the step of temperature compensation of the initial alcohol content value based on temperature data to obtain the compensated alcohol content value includes: using a deep residual network model with embedded physical prior information to perform temperature compensation on the initial alcohol content value to obtain the compensated alcohol content value; the physical prior information includes temperature, initial alcohol content value, and cross terms and / or polynomial terms constructed from the two.

[0009] Preferably, the training process of the deep residual network model includes: Obtain a triplet dataset containing temperature, initial alcohol content, and actual compensated alcohol content as the training set; Using mean squared error as the loss function and minimizing the difference between the model's predicted values ​​and the true values ​​as the objective, the model parameters are optimized using the Adam or AdamW optimization algorithm.

[0010] Preferably, the method for establishing the liquid level-flow rate calibration curve includes: With the detection device initially empty, calibration liquid is pumped into the detection device at different flow rates; While the calibration liquid is being pumped into the detection device, the calibration liquid also flows out of the detection device to simulate the actual wine-making process; after the liquid level in the detection device stabilizes, multiple sets of stable liquid level data and corresponding flow rate data are recorded. Based on the recorded multiple sets of data points, curve fitting was performed to establish a mapping relationship curve between liquid level data and flow rate data.

[0011] Preferably, the preset transition conditions include a combination of multiple unit conditions or a single unit condition; Each unit condition includes a condition object, a condition comparison operator, and a corresponding condition threshold; The condition object is any one of the data collected in real time or the data calculated.

[0012] Preferably, the buoyancy calibration value is the buoyancy received by the float in the detection device after it is completely submerged in the calibration liquid; The liquid level calibration value is the height of the calibration liquid in the detection device corresponding to the buoyancy calibration value obtained by measurement.

[0013] An automatic wine extraction control system includes a host computer and a slave computer connected by communication; the slave computer includes a control module and a detection device and an actuator connected to the control module. The detection device includes a buoyancy acquisition module, a temperature acquisition module, and a liquid level acquisition module; The control module processes the data collected by the detection device and compares it with preset transition conditions. The control actuator guides the flowing wine in the detection device to the container of the corresponding wine segment.

[0014] Preferably, the host computer includes: The data display module shows the data collected in real time from the detection device and the data obtained after calculation and processing. The mode setting module allows for the pre-setting or modification of transition conditions; The data calibration module calibrates the calibration parameters for calculating the flow density of the liquor, and also calibrates the level-flow calibration curve.

[0015] Preferably, the buoyancy acquisition module includes a float suspended in the detection device. The float is connected to the lower end of a connector, and the upper end of the connector is connected to a fixing member on which a force sensor is installed. The force sensor detects buoyancy data by measuring the force changes of the float when it is not in contact with liquid and when it is completely submerged in liquid.

[0016] This invention offers the following advantages: Fully automated, high-precision measurement: Through multi-sensor fusion and advanced algorithm models (such as deep residual networks), it achieves automated, high-precision measurement of alcohol content, effectively eliminating temperature effects and hardware errors, with accuracy far exceeding manual and traditional methods. Intelligent, flexible transition: Users can flexibly set diverse transition conditions (such as a combination of alcohol content and cumulative mass judgment) according to the different process requirements of various wines, greatly improving the system's adaptability and reliability, and avoiding misjudgments caused by single signal distortion. High stability and reliability: Adopting a separate upper and lower computer architecture, even if the upper computer fails, the lower computer can still independently complete the wine extraction control, ensuring the continuity and stability of the production process. Increased production efficiency: Completely replacing manual labor, it significantly reduces labor intensity and improves wine extraction efficiency and product quality consistency. Attached Figure Description

[0017] Figure 1 This is a flowchart of an automatic wine extraction control method according to the present invention.

[0018] Figure 2 This is a schematic diagram of an automatic wine-picking control system according to the present invention. Detailed Implementation

[0019] The present invention will now be further described with reference to the accompanying drawings and specific embodiments.

[0020] like Figure 1 As shown, an automatic wine-picking control method includes: Real-time data collection of buoyancy, temperature, and liquid level of the flowing wine into the detection device; Based on buoyancy data and liquid level data, combined with preset calibration parameters, the real-time density of the flowing wine is calculated, and the initial alcohol content value is obtained according to the mapping relationship between density and alcohol content. The initial alcohol content value is compensated for based on temperature data to obtain the compensated alcohol content value. Based on liquid level data, the real-time volumetric flow rate is obtained according to the preset liquid level-flow rate calibration curve, and the mass flow rate is calculated. At least one of the real-time collected data and the calculated data is used as the judgment object and compared with the preset transition conditions to guide the flowing wine in the detection device to the container of the corresponding wine segment.

[0021] It should be noted that, through real-time monitoring by multiple sensors and corresponding measurement structure design, combined with pre-set judgment conditions for different liquor segments, this invention can extract liquor from different segments and calculate the cumulative mass and overall alcohol content of each segment, thereby improving the extraction efficiency and the stability and reliability of the liquor output. For the initial alcohol content obtained by measuring density mapping, it is also compensated and corrected by the temperature of the liquor to obtain a more accurate real-time alcohol content value. At the same time, the method for obtaining the real-time flow rate of the liquor is optimized, which can obtain a more accurate flow rate result and further improve the accuracy of automatic liquor extraction.

[0022] It is worth noting that in the automatic wine-collecting control method of this invention, the generated wine first flows into the detection device. The buoyancy detection module, temperature detection module, and liquid level detection module in the detection device collect relevant data. Then, based on this data, the actual alcohol content, volumetric flow rate, and mass flow rate of the wine are calculated. The weight of the wine flowing into the corresponding container for a specific wine segment is calculated based on the product of flow rate and time. Based on these collected and calculated data, the data is compared with preset transfer conditions. If a preset transfer condition is met, the wine is transferred to the corresponding container, completing the automatic wine-collecting process. Through multi-sensor fusion and advanced algorithm models (such as deep residual networks), automatic and high-precision measurement of alcohol content is achieved, effectively eliminating the influence of temperature and hardware errors, with accuracy far exceeding that of manual and traditional methods. Intelligent flexible transfer: Users can flexibly set diverse transfer conditions (such as a combination of alcohol content and cumulative mass judgment) according to the different process requirements of various wines, greatly improving the adaptability and reliability of the system and avoiding misjudgments caused by single signal distortion.

[0023] As a specific example, the first step is to calculate the real-time density of the flowing wine. The accurate real-time density result of the flowing wine is the basis for obtaining the accurate actual alcohol content of the flowing wine.

[0024] Based on buoyancy data and liquid level data, the real-time density of the flowing liquid is calculated using preset calibration parameters. This includes: using the ratio of the buoyancy calibration value to the density of the calibration liquid plus the volume error compensation value as the denominator and the buoyancy data as the numerator to calculate the real-time density of the flowing liquid; the volume error compensation value is calculated by the difference between the liquid level data and the liquid level calibration value.

[0025] The buoyancy calibration value is the buoyancy force received by the float in the detection device after it is completely submerged in the calibration liquid; the liquid level calibration value is the liquid level height of the calibration liquid in the detection device when the buoyancy calibration value is measured.

[0026] It should be noted that in this invention, the float is suspended in the detection device by a connector. The float is connected to the lower end of the connector, and the upper end of the connector is connected to a fixing member on which a force sensor is installed. The force sensor detects buoyancy data by measuring the force changes of the float when it is not in contact with the liquid and when it is completely submerged in the liquid.

[0027] It's worth noting that the density calculation is based on an improved formula using Archimedes' principle and the buoyancy method. Ideally, when the float is completely submerged in the liquid, the buoyancy force F is the liquid density multiplied by the float's volume multiplied by g. If the float's volume is known precisely, the corresponding liquid density can be calculated. However, in actual detection devices, the connecting parts used to suspend the float are also partially submerged in the liquid, generating additional buoyancy. The volume error compensation value is used to eliminate the volume error introduced by the connecting parts. Specifically, the difference between the real-time detected liquid level data and the calibrated liquid level value is the difference in the height of the connecting parts submerged in the liquid between the two buoyancy measurements. Multiplying this height difference by the cross-sectional area of ​​the connecting parts yields the volume error compensation value.

[0028] The ratio of the buoyancy calibration value to the density of the calibration liquid represents the volume of the float itself under calibration conditions. The ratio of the buoyancy calibration value to the density of the calibration liquid in the denominator, plus the volume error compensation value, can be equivalent to the total volume of liquid displaced by the float under the current conditions. The numerator is the real-time buoyancy data. This can compensate for errors introduced by the hardware structure of density measurement, resulting in a more accurate real-time density result for the flowing liquid.

[0029] As a specific implementation, after obtaining the real-time density of the flowing liquor, the actual alcohol content of the flowing liquor is acquired. Specifically, after calculating the real-time density, a lookup table method is used to look up the density and alcohol content against a pre-stored alcohol density concentration and alcohol content comparison table (which can be an existing standard parameter table or a table compiled based on historical measurement data). This table uses the density value as the index and the corresponding alcohol content as the value. By looking up the table or performing linear interpolation on adjacent entries, the initial alcohol content value corresponding to the current density is quickly obtained.

[0030] The initial alcohol content value is merely the raw data and does not account for the actual effect of temperature. In fact, the difference between high and low temperatures has a significant impact on alcohol content, sometimes by as much as 10 degrees. Therefore, it is necessary to compensate and correct the initial alcohol content value of the flowing wine based on the detected temperature to obtain a more accurate result.

[0031] The initial alcohol content value is compensated for temperature data to obtain the compensated alcohol content value, which includes: A deep residual network model with embedded physical prior information is used to perform temperature compensation on the initial alcohol content value to obtain the compensated alcohol content value; the physical prior information includes temperature, initial alcohol content value, and cross terms and / or polynomial terms constructed from the two.

[0032] It's important to note that traditional deep networks may encounter vanishing or exploding gradients as the number of layers increases, leading to training difficulties and deteriorating performance. Deep residual networks solve this problem by introducing "shortcut connections," allowing the network to learn "residuals." This means that a portion of the network can directly learn identity mappings, enabling the model to easily become very deep and thus capable of capturing the extremely complex nonlinear relationship between temperature and raw alcohol content, far exceeding the accuracy of linear interpolation.

[0033] However, when embedding physical prior information, the effects of initial alcohol content and temperature on the final result are not isolated. For example, the rate of change of alcohol content may be greater within a certain temperature range. An embedding network can be used to map the original input features to a higher-dimensional feature space. This process can explicitly include some interaction features (such as the product of temperature and initial alcohol content, interaction terms or polynomial terms with both as variables), making it easier for the main network to learn key patterns.

[0034] Specifically, the embedding of physical prior information serves as the model input. This includes not only the initial alcohol content and temperature, but also additional higher-order terms, interaction terms, and polynomial terms manually constructed based on the initial alcohol content and temperature for feature enhancement. These enhanced feature vectors are then input into the subsequent network structure. These higher-order terms and interaction terms help the model more easily learn the complex nonlinear interaction between temperature and alcohol content.

[0035] In terms of network structure, it includes an embedding layer, a residual block combination, and an output layer.

[0036] The embedding layer is a shallow, fully connected neural network. After receiving the augmented feature vector, it transforms it through one or two fully connected layers to obtain the embedded feature vector. The embedded feature vector can be obtained by combining the augmented feature vector with the learnable parameters of the embedding network and then using the nonlinear activation function ReLU.

[0037] Residual block assemblies are composed of stacked residual blocks and form the core of the entire network. Each residual block contains two or more fully connected layers, and its core feature is the shortcut connection, which directly adds the block's input to its output. For a residual block with input Zin, its output Zout can be represented as the sum of Zin calculated through the transformation function F and Zin itself. The transformation function F refers to the transformation function within the residual block, typically consisting of two fully connected layers and an activation function. This structure greatly alleviates the vanishing gradient problem in deep networks, enabling the construction of very deep networks to capture extremely complex nonlinear relationships, with fitting capabilities far exceeding any linear interpolation method.

[0038] In the output layer, the output of the last residual block in the residual block combination is mapped from the high-dimensional features to a single output value, namely the final compensated alcohol content value, through a fully connected layer.

[0039] Furthermore, the training process of the deep residual network model includes: Obtain a triplet dataset containing temperature, initial alcohol content, and actual compensated alcohol content as the training set; Using mean squared error as the loss function and minimizing the difference between the model's predicted values ​​and the true values ​​as the objective, the model parameters are optimized using the Adam or AdamW optimization algorithm.

[0040] It should be noted that the triplet dataset can be obtained by generating a large-scale training dataset based on the data in the national standard. The national standard provides correction values ​​for different temperatures and alcohol strengths. Therefore, countless triples (T, initial alcohol strength, compensated alcohol strength) can be generated, where the compensated alcohol strength is the true alcohol strength corresponding to the actual density at that temperature T as specified in the national standard. The dataset should cover all possible temperature and alcohol strength ranges.

[0041] It's worth noting that deep residual structures can effectively learn highly complex nonlinear patterns in data, and their fitting ability far surpasses that of linear interpolation. Through data augmentation and learning deep features, the model exhibits better robustness to minor noise or perturbations in the input data. If more influencing factors (such as pressure, other liquid components, etc.) need to be considered in the future, they can simply be added as new input features to the model without refactoring the entire algorithm. After training, the model provides a direct mapping from (temperature, initial alcohol content) to the compensated alcohol content, making it simple and efficient to run.

[0042] As a specific embodiment, after completing the actual alcohol content measurement of the liquor, it is also necessary to measure the flow rate data. The accuracy of the flow rate measurement directly affects the measurement of the cumulative mass. This invention abandons the traditional theoretical formula and adopts a calibration method based on actual measurement.

[0043] Methods for establishing level-flow calibration curves include: With the detection device initially empty, calibration liquid is pumped into the detection device at different flow rates; While the calibration liquid is being pumped into the detection device, the calibration liquid also flows out of the detection device to simulate the actual wine-making process; after the liquid level in the detection device stabilizes, multiple sets of stable liquid level data and corresponding flow rate data are recorded. Based on the recorded multiple sets of data points, curve fitting was performed to establish a mapping relationship curve between liquid level data and flow rate data.

[0044] Specifically, a high-precision electronic flow meter is installed on the outlet pipe of the testing device. The outlet valve of the testing device is kept at the same opening degree as during actual wine extraction. A calibration liquid (e.g., water) is pumped into the testing device at different constant flow rates Qi using a variable frequency water pump. For each flow rate Qi, the system is allowed to reach a steady state (inflow equals outflow, and the liquid level Hi in the device no longer changes). The flow meter reading Qi and the liquid level data Hi are recorded at this point. This process is repeated, from minimum to maximum flow rate, collecting several sets of data points (Hi, Qi). The collected data points are plotted on a coordinate system. The HQ relationship is usually not strictly linear, but rather a curve. A least-squares method can be used for polynomial fitting (e.g., it can be fitted to a quadratic polynomial, power function, or other function based on worker experience or directly calculated by computer software). The fitted curve equation is the liquid level-flow rate calibration curve.

[0045] After calibrating the level-flow rate calibration curve, during real-time flow rate detection, the real-time liquid level height within the detection device is read and substituted into the level-flow rate calibration curve to calculate the corresponding volumetric flow rate. This method minimizes the influence of the detection device's geometry, manufacturing precision, and installation errors on flow rate measurement, resulting in extremely high accuracy. After obtaining the volumetric flow rate, the corresponding mass flow rate is obtained by multiplying it by the previously detected real-time density of the flowing liquor. The mass flow rate of each liquor segment is then accumulated over time to obtain the cumulative mass of that segment.

[0046] As a specific embodiment, the preset transition conditions include a combination of multiple unit conditions or a single unit condition; each unit condition includes a condition object, a condition comparison operator, and a corresponding condition threshold. The condition object is any one of the data collected in real time or the data calculated.

[0047] It's important to note that the entire distillation process is generally divided into five stages: first distillate, second distillate, third distillate, final distillate, and tail liquor. The quality of the liquor varies between these stages, requiring different liquors to be stored in separate tanks. Therefore, automatic stage transition and separation of liquors from different stages are necessary. First, the system (P) acquires pre-defined mode data, including information required for the stage transition logic: transition condition objects (such as alcohol content, distillation time, temperature, etc.), the values ​​of these condition objects, and the comparison methods for each condition (greater than, less than, equal to, etc.). After distillation begins, the first distillate stage begins. Subsequently, real-time values ​​are compared with the values ​​of the corresponding objects in the transition conditions. If the transition conditions are met, the system proceeds to the next stage; otherwise, it continues operating on the current stage until the tail liquor stage, the final stage. After all liquid has flowed, the system returns to its initial state and repeats the above steps.

[0048] The specific transition control process is as follows: Condition Settings: Conditions for transitioning from one distillation stage to the next can be set in advance for each stage. For example: First Distillation -> Second Distillation: The condition can be set to [Compensated Alcohol Content ≤ 68%] AND [Cumulative Mass > 2.5kg]. This means that even if the alcohol content quickly drops below 68%, the second distillation will only begin collecting after the cumulative mass of the first distillation reaches 2.5kg, ensuring complete separation of the first distillation. Second Distillation -> Third Distillation: The condition can be set to [Compensated Alcohol Content ≤ 60.0%.] This is a single-condition judgment; as long as the alcohol content is less than or equal to 60%, the distillation will transition to the third distillation stage. Third Distillation -> Tail Distillation: The condition can be set to [Compensated Alcohol Content ≤ 52.0%] OR [Drinking Time > 3600 seconds]. This is an OR logic to prevent the distillation from failing due to a malfunctioning alcohol content sensor. If the timeout occurs, the distillation will be forcibly switched to tail distillation collection, ensuring production safety.

[0049] Throughout the entire automated wine extraction process, the values ​​of all conditional objects are monitored in real time. At each interval, the system automatically checks whether the preset transition conditions are met. A transition action is triggered only when all conditions set to "AND" are met, or when one of the conditions set to "OR" is met. After triggering the transition, the control module sends corresponding control commands to the actuator. The actuator precisely rotates to the next predetermined angle at a preset speed and acceleration, driving the diversion valve to switch the outlet and introduce the wine into a new container. Simultaneously, the counters for the cumulative mass and flow time of the new wine segment are reset to zero, and a new accumulation process begins.

[0050] In addition to an automatic wine-collecting control method, this invention also provides, for example, Figure 2 The automatic wine-picking control system shown includes an upper computer and a lower computer connected by communication; the lower computer includes a control module and a detection device and an actuator connected to the control module. The detection device includes a buoyancy acquisition module, a temperature acquisition module, and a liquid level acquisition module; The control module processes the data collected by the detection device and compares it with the preset transition conditions. The control actuator then guides the flowing wine in the detection device to the container of the corresponding wine segment.

[0051] It is worth noting that this invention adopts a separate upper and lower computer architecture. Even if the upper computer fails, the lower computer can still independently complete the wine extraction control, ensuring the continuity and stability of the production process. The upper computer is developed using SCADA configuration software, including a data display module, a mode setting module, and a data calibration module. The lower computer includes a control module and detection devices and actuators interconnected with the control module. The detection devices are equipped with temperature acquisition modules, liquid level acquisition modules, and buoyancy acquisition modules. Each acquisition module has corresponding sensors. The control module mainly uses a PLC to drive the rotation of a servo motor through a servo drive controller. The upper computer communicates with the PLC through OPCUA. The PLC program controls the servo drive through the ETHERCAT bus to rotate the motor in the actuator to the corresponding position, that is, the wine flow pipe points to the container of the corresponding wine segment.

[0052] As a specific embodiment, the host computer can be an industrial computer, running a monitoring interface developed based on SCADA configuration software, and communicating with the PLC via the OPC UA protocol. It includes: The data display module shows the data collected in real time from the detection device and the data obtained after calculation and processing. The mode setting module allows for the pre-setting or modification of transition conditions; The data calibration module calibrates the calibration parameters for calculating the flow density of the liquor, and also calibrates the level-flow calibration curve.

[0053] The data display module shows all key data in real time, including compensated alcohol content, real-time temperature, real-time flow rate, brewing time, cumulative quality, and current brewing stage, in the form of numbers, trend curves, and dashboards, providing operators with intuitive production status monitoring. The PLC transmits data to the pre-planned corresponding tag number to the host computer via OPCUA. The host computer acts as the OPCUA client, and the PLC acts as the server. The host computer reads the data every 1 second and displays it on the screen in real time.

[0054] The mode setting module includes segment settings, condition object settings (selectable alcohol content, flow rate, and flow time), condition settings (>, ≥, <, ≤, =, ≠), and condition threshold settings. After these data are set, the data is written to the pre-planned OPCUA communication tag number via button click events and then transmitted to the PLC via OPCUA. It provides a flexible segment transition condition configuration interface. Users can select "Condition Object" and "Comparison Operator" from the drop-down menu and enter the "Condition Threshold." Multiple unit condition logic combinations (such as "AND" logic) can be set. All setting parameters are written to the PLC's memory area via OPCUA.

[0055] The data calibration module includes calibration of liquid level, buoyancy, and liquid level-flow rate calibration curves. These three calibrations are to eliminate some errors in the hardware structure manufacturing of the detection device and avoid affecting the alcohol content measurement results.

[0056] In one specific embodiment, the lower-level machine includes a control module and detection devices and actuators interconnected with the control module. The detection device includes a buoyancy acquisition module, a liquid level acquisition module, and a temperature acquisition module. The control module is responsible for data acquisition from all sensors, real-time operation of all core algorithms (density, alcohol content, and flow rate calculations), judgment of transition logic, and ultimately, precise control of the actuators.

[0057] The buoyancy acquisition module includes a float suspended within the detection device. The float is connected to the lower end of a connector, and the upper end of the connector is connected to a fixing component housing a force sensor. The force sensor detects buoyancy data by measuring the force changes experienced by the float when it is not in contact with liquid and when it is completely submerged. Initially, the weight G0 of the float in air is measured. As liquid flows into the device and the float gradually becomes submerged, according to Archimedes' principle, the float experiences a buoyant force F. At this point, the force value Fsensor measured by the force sensor becomes G0 minus F. Therefore, the buoyant force F is G0 minus Fsensor. This measurement method avoids directly measuring buoyancy, instead measuring the change in force, resulting in higher accuracy and stronger anti-interference capabilities.

[0058] The temperature acquisition module can use a PT100 platinum resistance temperature sensor, which has the advantages of high measurement accuracy, good stability, and excellent linearity. The probe of the temperature sensor is positioned so that it can be directly immersed in the flowing wine, usually installed on the side wall of the detection device or at the inlet pipe, ensuring that it can quickly and accurately reflect the true temperature of the flowing wine.

[0059] The liquid level acquisition module can use a hydrostatic level sensor or an ultrasonic level gauge. The hydrostatic sensor calculates the liquid level by measuring the static pressure of the liquid and is installed at the bottom of the detection device; the ultrasonic level gauge is installed at the top and provides non-contact measurement.

[0060] The actuator is responsible for completing the final dispensing of the liquor. The PLC communicates with the servo drive using the ETHERCAT communication protocol. The motor is controlled by the segmentation logic and the CIA402 motion function library to rotate to the designated position, ensuring the dispensing spout is accurately aligned with the different containers. In each segment, the motor positions itself accordingly, such as container number 1; subsequently, in the next segment, the motor adjusts to the next position, such as container number 2, and so on. This method distinguishes liquors of different strengths, thus achieving automatic dispensing.

[0061] The above embodiments are further elaborations and descriptions of the present invention to facilitate understanding, and are not intended to limit the present invention in any way. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An automatic wine-collecting control method, characterized in that, include: Real-time data collection of buoyancy, temperature, and liquid level of the flowing wine into the detection device; Based on buoyancy data and liquid level data, combined with preset calibration parameters, the real-time density of the flowing wine is calculated, and the initial alcohol content value is obtained according to the mapping relationship between density and alcohol content; the initial alcohol content value is compensated for by temperature data to obtain the compensated alcohol content value. Based on the liquid level data, the real-time volumetric flow rate is obtained according to the preset liquid level-flow rate calibration curve, and the mass flow rate is calculated. At least one of the real-time collected data and the calculated data is used as the judgment object and compared with the preset transition conditions to guide the flowing wine in the detection device to the container of the corresponding wine segment.

2. The automatic wine extraction control method according to claim 1, characterized in that, The calculation of the real-time density of the flowing wine based on buoyancy data and liquid level data, combined with preset calibration parameters, includes: The real-time density of the liquid is calculated by adding the volume error compensation value to the ratio of the buoyancy calibration value to the density of the calibration liquid, and using the buoyancy data as the numerator. The volume error compensation value is calculated by the difference between the liquid level data and the liquid level calibration value.

3. The automatic wine-collecting control method according to claim 1 or 2, characterized in that, The process of performing temperature compensation on the initial alcohol content value based on temperature data to obtain the compensated alcohol content value includes: A deep residual network model with embedded physical prior information is used to perform temperature compensation on the initial alcohol content value to obtain the compensated alcohol content value; the physical prior information includes temperature, initial alcohol content value, and cross terms and / or polynomial terms constructed from the two.

4. The automatic wine extraction control method according to claim 3, characterized in that, The training process of the deep residual network model includes: Obtain a triplet dataset containing temperature, initial alcohol content, and actual compensated alcohol content as the training set; Using mean squared error as the loss function and minimizing the difference between the model's predicted values ​​and the true values ​​as the objective, the model parameters are optimized using the Adam or AdamW optimization algorithm.

5. An automatic wine-collecting control method according to claim 1, 2, or 4, characterized in that, The method for establishing the liquid level-flow rate calibration curve includes: With the detection device initially empty, calibration liquid is pumped into the detection device at different flow rates; The testing device pumps in calibration liquid while simultaneously releasing calibration liquid from the device to simulate the actual wine-making process. After the liquid level in the detection device stabilizes, record multiple sets of stable liquid level data and corresponding flow rate data; Based on the recorded multiple sets of data points, curve fitting was performed to establish a mapping relationship curve between liquid level data and flow rate data.

6. An automatic wine-collecting control method according to claim 1, 2, or 4, characterized in that, The preset transition conditions include a combination of multiple unit conditions or a single unit condition; Each unit condition includes a condition object, a condition comparison operator, and a corresponding condition threshold; The condition object is any one of the data collected in real time or the data calculated.

7. The automatic wine extraction control method according to claim 2, characterized in that, The buoyancy calibration value is the buoyancy force received by the float in the detection device after it is completely submerged in the calibration liquid; The liquid level calibration value is the height of the calibration liquid in the detection device corresponding to the buoyancy calibration value obtained by measurement.

8. An automatic wine-collecting control system, applicable to the automatic wine-collecting control method as described in any one of claims 1-7, characterized in that, It includes a host computer and a slave computer connected by communication; the slave computer includes a control module and a detection device and an execution mechanism interconnected with the control module. The detection device includes a buoyancy acquisition module, a temperature acquisition module, and a liquid level acquisition module; The control module processes the data collected by the detection device and compares it with preset transition conditions. The control actuator guides the flowing wine in the detection device to the container of the corresponding wine segment.

9. An automatic wine-collecting control system according to claim 8, characterized in that, The host computer includes: The data display module shows the data collected in real time from the detection device and the data obtained after calculation and processing. The mode setting module allows for the pre-setting or modification of transition conditions; The data calibration module calibrates the calibration parameters for calculating the flow density of the liquor, and also calibrates the level-flow calibration curve.

10. An automatic wine-collecting control system according to claim 8 or 9, characterized in that, The buoyancy acquisition module includes a float suspended inside the detection device. The float is connected to the lower end of a connector, and the upper end of the connector is connected to a fixing member on which a force sensor is installed. The force sensor detects buoyancy data by measuring the force changes of the float when it is not in contact with liquid and when it is completely submerged in liquid.

Images