PU (Poly Urethane) value online monitoring and intelligent adjusting system in wine sterilization process

Abstract

The invention discloses a PU value online monitoring and intelligent adjusting system in a wine sterilization process, and the system comprises an identification loading module which is used for reading container identification information before a container to be sterilized enters a sterilization device, and loading a product thermosensitive parameter set corresponding to the container according to the container identification information; the multi-source acquisition module is used for acquiring the temperature and the conveying speed of a spraying medium according to spraying partitions in the sterilization process, and synchronously acquiring the temperature of the outer surface of the container to form a continuous temperature sequence; and the reference temperature measurement module is used for selecting an accompanying reference container and acquiring an internal temperature sequence of the accompanying reference container for constructing an internal temperature calibration constraint. According to the invention, through combination of cold point sensing, PU conservative calculation and target adaptive adjustment, precise control of sterilization intensity and thermal load minimization are realized.

Description

Technical Field

[0001] This application relates to the field of alcohol sterilization control system technology, and in particular to an online monitoring and intelligent adjustment system for the PU value of alcohol sterilization process. Background Technology

[0002] After bottling and capping, beer (especially craft beer) is often pasteurized using tunnel spray. The pasteurization intensity is characterized by the PU value, which aims to minimize flavor aging, foam reduction, and color darkening caused by heat load while ensuring biological stability. In the industry, because the temperature inside the bottle or can is difficult to measure directly and continuously, the PU is often estimated based on the spray water temperature and empirical models, and adjustments are made accordingly.

[0003] Existing technologies mostly focus on using temperature control and models to stabilize the "final PU" at a set value, or determining the minimum PU value offline through sample retention and microbial testing. However, there are still two easily overlooked problems in craft brewing with multiple SKUs, fluctuating production cycles, and frequent line changes: First, the "cold spot" inside the container can drift with viscosity / turbidity, gas evolution, headspace height, tank type, and wall thickness. The fixed structure of the accompanying recorder and probe may also introduce thermal bridges and thermal artifacts, causing long-term systematic deviations in online PU calculations, resulting in a small number of containers being under-sterilized or the entire container being over-sterilized for safety. Second, the differences in microbial risk, heat resistance, and quality sensitivity of different formulations invalidate the implicit premise that "the same PU = the same sterilization effect," and fixing the PU target may be neither optimal nor safe. Summary of the Invention

[0004] To address the aforementioned problems, embodiments of the present invention provide an online monitoring and intelligent adjustment system for the PU value during the alcohol sterilization process. The system includes:

[0005] Identification and loading module: Before the container to be sterilized enters the sterilization device, it reads the container identification information and loads the product thermal parameter set corresponding to the container accordingly;

[0006] Multi-source acquisition module: During the sterilization process, the temperature of the spray medium and the conveying speed are collected according to the spray zone, and the temperature of the outer surface of the container is collected simultaneously to form a continuous temperature sequence;

[0007] Reference temperature measurement module: Selects the accompanying reference container and obtains its internal temperature sequence, which is used to construct internal temperature calibration constraints;

[0008] Heat transfer modeling module: Based on the spray medium temperature, conveying speed and the product's thermosensitive parameter set, establish a heat transfer state model that can describe the internal candidate cold points;

[0009] Fusion estimation module: Input the continuous temperature sequence of the outer surface of the container and the internal temperature sequence of the reference container into the fusion estimation algorithm to correct the heat transfer state model online and obtain the internal cold point temperature trajectory and its uncertainty range;

[0010] Cold spot reconstruction module: By comparing the deviation between the internal temperature sequence of the reference container and the prediction results of each candidate cold spot, the internal cold spot location is reconstructed in real time and the heat transfer state model is updated.

[0011] PU calculation module: Based on the uncertainty range of the cold spot temperature trajectory, a conservative temperature trajectory that is not conducive to sterilization is selected and time integration is performed to obtain the real-time PU value;

[0012] Target generation module: Generates dynamic target PU values ​​based on the product's thermal parameter set and the device's operational risk status;

[0013] Joint adjustment module: compares the real-time PU value with the target PU value, generates and executes joint adjustment commands for the spray medium temperature and conveying speed.

[0014] Furthermore, the fusion estimation algorithm includes: performing first-order dynamic filtering on the temperature sequence of the outer surface of the container to obtain a denoised temperature response; performing time alignment processing on the temperature sequence of the reference container; and adjusting the heat transfer gain parameters in the heat transfer state model online based on the instantaneous deviation between the predicted internal temperature of the outer surface temperature and the internal temperature of the reference container within each control cycle, so that the predicted internal cold point temperature gradually converges to the reference constraint range.

[0015] Furthermore, the process of reconstructing the location of the internal cold spot includes:

[0016] In the heat transfer state model, N candidate cold point positions are discretized along the axial and radial directions of the container; the internal temperature prediction trajectory is calculated for each candidate cold point; the cumulative residual between the predicted trajectory and the internal temperature sequence of the reference container is calculated within a time window; the candidate cold point with the smallest cumulative residual is selected as the actual cold point under the current sterilization condition, and this cold point is used for subsequent PU calculations.

[0017] Furthermore, the method for selecting the conservative temperature trajectory includes:

[0018] Within the uncertainty range corresponding to the internal cold point temperature trajectory, the lower limit of the temperature is selected as the input for calculating the sterilization effect; and when the conveying speed changes or the spray medium temperature changes abruptly, the uncertainty range is expanded to improve the safety margin of the real-time PU value under dynamic operating conditions.

[0019] Furthermore, the method for generating the dynamic target PU value includes:

[0020] The product's thermal parameters are mapped to the maximum allowable cumulative heat load; the device's operating status is mapped to the sterilization risk coefficient; the adjustment direction and magnitude of the target PU value are determined jointly based on the upper limit of the heat load and the sterilization risk coefficient, so that the target PU value meets the sterilization requirements while avoiding unnecessary heat treatment.

[0021] Furthermore, the joint adjustment command is generated by a rolling optimization control algorithm, which in each control cycle: predicts the PU change trend in the future time step; uses the deviation between the real-time PU value and the target PU value as a constraint; and, under the premise of satisfying the PU constraint, prioritizes the control action that has less impact on the product's heat load, either the spray medium temperature adjustment or the conveying speed adjustment.

[0022] Furthermore, when a change in container identification information is detected, a product switching process is executed, including: recording the thermal state of each spray zone at the moment of switching; predicting the initial thermal environment when the first batch of new product containers enter each spray zone; and, based on the prediction, making advance correction to the spray medium temperature setpoint to ensure a continuous and smooth PU accumulation process for the first batch of new product containers.

[0023] Furthermore, after the batch is completed, the cold spot temperature trajectory recorded within the batch and the corresponding PU calculation results are statistically analyzed. When a systematic trend of PU deviation is detected in consecutive batches, the model parameters related to spray heat exchange in the heat transfer state model are automatically corrected.

[0024] An online monitoring and intelligent adjustment system for PU value in the sterilization process of alcoholic beverages also includes: a model verification module: after the batch ends, a model consistency verification and parameter update triggering mechanism for cross-batch operation is introduced to determine whether the current heat transfer state model is still applicable to subsequent batch operation, and adaptively correct the model when the update conditions are met, and keep the model parameters frozen when the conditions are not met.

[0025] The technical effects and advantages of the online monitoring and intelligent adjustment system for PU value in the wine sterilization process provided by this invention are as follows:

[0026] This invention achieves precise control of sterilization intensity and minimizes heat load by combining cold spot sensing, conservative PU calculation, and adaptive target adjustment. It utilizes multi-source data acquisition, reference temperature measurement, and fusion estimation to correct the heat transfer state model online, dynamically identify and track internal cold spot locations, effectively reducing systematic errors in the PU caused by cold spot drift. By employing a conservative temperature trajectory based on uncertainty ranges for PU calculation, it automatically increases safety margins during operating condition fluctuations, reducing the risk of under-sterilization while avoiding long-term over-sterilization. It incorporates the product's thermosensitive characteristics and the device's operational risks into the target PU generation process, enabling the sterilization target to adapt to product differences and operating conditions, balancing biological stability and quality protection. Through rolling optimization, it selects control actions with minimal heat load impact from various adjustment methods, achieving coordinated adjustment of spray temperature and conveying cycle time, improving system stability under conditions of multiple SKUs and frequent line changes. Finally, it introduces a cross-batch model verification and self-correction mechanism, allowing the heat transfer model to continuously calibrate with long-term operating conditions, enhancing the long-term consistency and reliability of online PU monitoring and adjustment. Attached Figure Description

[0027] Figure 1 This is a schematic diagram showing the connection of an online monitoring and intelligent adjustment system for the PU value of alcoholic beverage sterilization process in Example 1.

[0028] Figure 2 This is a schematic diagram of the connection of an online monitoring and intelligent adjustment system for the PU value of alcoholic beverages in Example 2. Detailed Implementation

[0029] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] Example 1:

[0031] Please see Figure 1 As shown, an embodiment of the present invention provides an online monitoring and intelligent adjustment system for the PU value of alcoholic beverages during the sterilization process, comprising:

[0032] Identification and loading module: Before the container to be sterilized enters the sterilization device, it reads the container identification information and loads the product thermal parameter set corresponding to the container accordingly;

[0033] Multi-source acquisition module: During the sterilization process, the temperature of the spray medium and the conveying speed are collected according to the spray zone, and the temperature of the outer surface of the container is collected simultaneously to form a continuous temperature sequence;

[0034] Reference temperature measurement module: Selects the accompanying reference container and obtains its internal temperature sequence, which is used to construct internal temperature calibration constraints;

[0035] Heat transfer modeling module: Based on the spray medium temperature, conveying speed and the product's thermosensitive parameter set, establish a heat transfer state model that can describe internal candidate cold points; internal cold points refer to the parts of the object being treated (e.g., materials inside a container, the inside of a package or the inside of a product) with the lowest temperature at any given time and which are most unfavorable for achieving sterilization standards. Internal cold points may not necessarily be directly measured by sensors and are often inferred through external temperature measurement, process parameters and heat transfer models.

[0036] Fusion estimation module: Input the continuous temperature sequence of the outer surface of the container and the internal temperature sequence of the reference container into the fusion estimation algorithm to correct the heat transfer state model online, and obtain the internal cold point temperature trajectory and its uncertainty interval. The internal cold point temperature trajectory is the temperature curve of this part changing with time, which is usually inferred from external temperature measurement data, transport speed and heat transfer relationship. Since the inference process is inevitably affected by measurement error, model simplification and dynamic response lag, the system constructs an uncertainty interval around the temperature trajectory to describe the possible upper and lower ranges of the actual internal cold point temperature.

[0037] Cold spot reconstruction module: By comparing the deviation between the internal temperature sequence of the reference container and the prediction results of each candidate cold spot, the internal cold spot location is reconstructed in real time and the heat transfer state model is updated.

[0038] PU calculation module: Based on the uncertainty range of the cold spot temperature trajectory, a conservative temperature trajectory that is not conducive to sterilization is selected and time integration is performed to obtain the real-time PU value;

[0039] Target generation module: Generates dynamic target PU values ​​based on the product's thermal parameter set and the device's operational risk status;

[0040] Joint adjustment module: compares the real-time PU value with the target PU value, generates and executes joint adjustment commands for the spray medium temperature and conveying speed.

[0041] In the fusion estimation module of this embodiment, the fusion estimation algorithm is used to correct the internal cold point temperature of the container online during the sterilization process. Its core purpose is to keep the internal cold point temperature predicted by the heat transfer state model consistent with the actual internal thermal state without directly measuring the internal temperature of each container to be sterilized.

[0042] Fusion estimation algorithms include:

[0043] Let the extended state be: ,in Indicates the internal thermal state of the container. The heat transfer gain parameter in the heat transfer state model represents the discrete time step. A set of fusion estimates is recursively given by:

[0044] predict: ; ;

[0045] renew:

[0046] ;

[0047] ;

[0048] ;

[0049] The measurement vector is taken as: ;

[0050] In the formula, For at any time Prior estimates, using Information at time −1 was predicted; For at any time The posterior estimate is obtained by incorporating measurements from this period; and These correspond to the prior and posterior estimation error covariances, respectively, and are used to quantify the uncertainty. The collected operating condition input vector is used to drive the heat transfer state model, typically including: spray medium temperature and delivery speed. The prediction function for the extended state represents "heat transfer state model + parameter evolution over time", used to map the previous period estimate and the current period input to the current period prediction; The measurement mapping function maps the extended state to the observable space, i.e., the output pairs. (External surface temperature) and (Predicted value of the internal temperature of the reference container); , This is the temperature sensitivity coefficient, representing the change in sterilization temperature. At ℃, the rate of heat-induced death of microorganisms changes tenfold. Here is the Jacobian matrix for the state, used to linearize the propagation uncertainty; , The Jacobian matrix for the state is used to linearize measurement updates; The two measurements are combined into a unified measurement vector to achieve "fusion"; The process noise covariance characterizes the uncertainty of model prediction error and parameter drift; To measure noise covariance and characterize the noise levels and reliability of external surface temperature measurements and reference container internal temperature measurements, the reliability of which measurement is greater is reflected in... Size; The Kalman gain matrix determines how the predicted and measured values ​​are weighted and fused within the current period. It is an identity matrix.

[0051] During the sterilization process, the outer surface temperature of the container is continuously collected by a temperature sensor installed above the conveyor line, forming a time-varying outer surface temperature sequence. Since the outer surface temperature signal is prone to high-frequency fluctuations due to spray impact, environmental disturbances, and sensor thermal inertia, this temperature sequence first enters a dynamic filtering process. This filtering process, constrained by time continuity, suppresses abrupt changes and noise, thereby obtaining a smooth outer surface temperature response that reflects the heating trend of the container, which is used for subsequent internal temperature inference.

[0052] Meanwhile, the reference container passes through the sterilization device along with the production line. The reference container is equipped with a temperature acquisition device to obtain the continuous change sequence of the internal temperature of the wine. Since there are differences between the internal temperature signal and the external surface temperature signal of the reference container in terms of sampling time and response speed, the internal temperature sequence of the reference container is time aligned before entering the fusion estimation to make it comparable with the external surface temperature sequence within the same control cycle. This alignment process does not change the temperature change trend, but is only used to eliminate the time offset caused by sampling delay.

[0053] After the above preprocessing is completed, the external surface temperature sequence and the internal temperature sequence of the reference container are used as observation inputs to enter the fusion estimation process. At this time, the heat transfer state model gives the predicted value of the internal cold point temperature based on the current spray medium temperature, conveying speed and container characteristics. The fusion estimation process calculates the instantaneous deviation between the predicted value of the internal temperature derived from the external surface temperature and the measured value of the internal temperature of the reference container in each control cycle.

[0054] When the deviation shows a stable direction within a continuous control cycle, it is considered that the heat transfer gain parameter in the current heat transfer state model does not match the actual operating conditions. At this time, the fusion estimation algorithm uses the deviation as the basis for correction and makes a small adjustment to the heat transfer gain parameter in the heat transfer state model, so that the model's description of the heat exchange intensity between the spray medium and the container is closer to the real process. This adjustment process is a gradual update to avoid drastic changes in the model due to fluctuations in a single measurement.

[0055] In the above manner, the heat transfer state model is continuously constrained by the internal temperature of the reference container during the sterilization process. Its output prediction of the internal cold point temperature gradually converges to the reasonable range defined by the internal temperature of the reference container. Since this correction process is continuously executed throughout the sterilization process, the model can still effectively track the internal cold point temperature even if the spraying conditions, conveying speed or environmental conditions change.

[0056] For example: In a certain implementation scenario, when the temperature of the spray medium rises briefly, the temperature change on the outer surface occurs before the temperature change inside the reference container. The fusion estimation process compares the magnitude of the changes of the two with the time relationship, and automatically reduces the growth rate of the corresponding heat transfer gain parameter in the model, so that the internal cold point temperature prediction is not overestimated due to the instantaneous temperature rise of the outer surface, thereby avoiding the real-time PU value being erroneously amplified.

[0057] During tunnel-type spray pasteurization, the lowest temperature zone inside the container is not always fixed in the same geometric position. Its position will shift with the temperature distribution of the spray medium, changes in the conveying speed, and the heating history of the container. If the internal cold point position is always assumed to remain unchanged, the temperature input used for PU calculation will be inconsistent with the actual weak sterilization area, thus introducing systematic errors. Therefore, the cold point reconstruction module in this embodiment reconstructs the internal cold point position online during the sterilization process.

[0058] In practice, when establishing the heat transfer state model, a unique internal cold point location is not preset. Instead, multiple candidate cold point locations are discretely set along the axial and radial directions inside the container. These candidate cold point locations represent several internal locations that may become the lowest temperature region under different operating conditions. Their number and distribution are determined by the spatial discretization method of the heat transfer state model and are used to cover the regions inside the container where temperature lag may occur.

[0059] During the sterilization process, the heat transfer state model calculates the corresponding internal temperature prediction trajectory for each candidate cold point location based on the spray medium temperature, conveying speed, and product thermal sensitivity parameters. The prediction trajectory describes the change of internal temperature over time under the current sterilization conditions, assuming that the candidate location is a cold point.

[0060] Meanwhile, the reference container passes through the sterilization device along the conveyor line, and its internal temperature sequence serves as the basis for observing the actual internal temperature changes. To avoid interference from single-point instantaneous fluctuations in the cold point judgment, the internal temperature sequence of the reference container is aligned with the predicted temperature trajectory of each candidate cold point within the same time window, and the cumulative residual between the two within the time window is calculated. The cumulative residual reflects the overall degree of deviation between the candidate cold point prediction result and the actual internal temperature change trend.

[0061] Within each reconstruction cycle, the cumulative residuals corresponding to all candidate cold point locations are compared, and the candidate cold point location with the smallest cumulative residual is selected as the actual internal cold point under the current sterilization conditions. This selection result indicates that, under the current spraying and conveying conditions, the temperature change at this candidate location best represents the true thermal response inside the reference container.

[0062] Once the location of the internal cold spot is determined, the heat transfer state model updates the corresponding description of the cold spot location. Subsequent internal temperature prediction, temperature uncertainty assessment, and PU calculation are all based on the reconstructed cold spot location, thereby ensuring that the real-time PU value always corresponds to the location that is most unfavorable for sterilization under the current operating conditions.

[0063] For example: After the conveying speed changes, the candidate cold spot near the bottom of the container shows a greater temperature lag in a short period of time. The cumulative residual between its predicted trajectory and the internal temperature of the reference container is significantly reduced. Based on this, the system automatically switches the internal cold spot position to the candidate position and uses the temperature trajectory corresponding to the position in the subsequent PU calculation, thereby avoiding underestimation of the sterilization risk due to improper assumption of the cold spot position.

[0064] In this embodiment, the PU calculation module is used to generate a conservative temperature trajectory for input calculation before calculating the bactericidal effect. The module is based on the internal cold point temperature trajectory, but does not directly use a single estimation result. Instead, it first constructs the corresponding uncertainty interval, then selects the lower limit of the temperature in the interval as the input for bactericidal effect calculation, and expands the uncertainty interval when dynamic operating conditions occur, thereby improving the safety margin of real-time PU value calculation at the mechanism level.

[0065] The uncertainty interval is formed by the synthesis of multiple types of interpretable errors: first, measurement errors caused by the accuracy of the sensors themselves and long-term drift, such as the temperature of the spray medium; second, model errors caused by the approximation of the internal cold point inference model in terms of physical properties and heat transfer conditions; third, time lag and position alignment errors in the internal temperature response of the product when the conveying speed changes or the spraying conditions are adjusted. The above errors are uniformly mapped into uncertainties in the temperature dimension, forming the upper and lower envelopes of the internal cold point temperature trajectory on the time axis.

[0066] In the PU calculation module, the system selects the lower limit of the temperature range as a conservative temperature trajectory and uses it as the input for the sterilization effect calculation (such as real-time PU value accumulation). Since the sterilization effect is highly sensitive to temperature, using the lower limit input can assume that the internal cold point is in a less favorable state at each moment, thereby avoiding the sterilization effect being overestimated due to overestimation.

[0067] When a change in conveying speed or a sudden change in spray medium temperature is detected, the operating condition is determined to have entered the dynamic stage. At this time, it is more difficult to accurately match the residence time of the product in each processing section and the internal temperature response, and the original uncertainty range is insufficient to cover the actual deviation. Therefore, the uncertainty range is temporarily expanded according to the preset logic, focusing on increasing the part related to time alignment error and dynamic lag, so that the upper and lower boundaries are widened, and the conservative temperature trajectory is further shifted downward accordingly. After the conveying speed or spray medium temperature stabilizes again and continues to meet the stability criterion, the uncertainty range gradually converges to the normal level to avoid unnecessary jumps in input temperature and real-time PU value.

[0068] For example: Under a certain stable operating condition, the center estimate of the internal cold point temperature is about 72℃, and the comprehensive uncertainty is ±1℃. The uncertainty range is 71℃~73℃. The PU calculation module selects 71℃ as the input for the sterilization effect calculation. When the conveying speed is significantly adjusted, the system expands the uncertainty to ±2℃, and the corresponding range becomes 70℃~74℃. The conservative input is then adjusted to 70℃, thereby automatically improving the safety margin of the real-time PU value calculation in the dynamic stage.

[0069] Through the above implementation method, the PU calculation module not only clarifies the selection basis of the conservative temperature trajectory, but also gives the specific triggering logic and mechanism of uncertainty range expansion and recovery under dynamic operating conditions, ensuring that the real-time sterilization effect assessment is neither overly conservative nor underestimates the risk in the critical transition stage.

[0070] In this embodiment, the target generation module is used to generate dynamic target PU values. Its function is to dynamically adjust the target PU values ​​based on the product's heat resistance and the device's operating status, while meeting the product's sterilization requirements, thereby avoiding unnecessary heat treatment due to excessive conservatism. This module works in conjunction with the PU calculation module: the PU calculation module focuses on the conservatism of real-time PU value calculation, while the target generation module focuses on the rationality and adaptability of the target value itself.

[0071] For ease of understanding, the relevant terms are explained as follows: The dynamic target PU value is a control target that is updated during operation as the operating conditions change. Its base value comes from the product's sterilization compliance requirements; the product's heat-sensitive parameters describe the product's tolerance to heat treatment and can be derived from process documents, quality verification data, or historical experience; the maximum allowable cumulative heat load is an upper limit index mapped from the product's heat-sensitive parameters, used to constrain the cumulative impact of heat treatment on product quality; the device operating status reflects whether the equipment and operating conditions are stable; and the sterilization risk coefficient is a quantitative result of the degree of sterilization uncertainty under the current operating conditions.

[0072] In the target generation module, the maximum cumulative heat load limit is first generated based on the product's thermal sensitivity parameters. This mapping is related to the product quality change mechanism. For example, the "overall thermal history that can be tolerated before unacceptable quality changes" is transformed into a heat load index that can be accumulated online. In implementation, the system uses the upper side or the center of the temperature trajectory with a conservative temperature trajectory in the uncertainty range of the internal cold point temperature trajectory to update the cumulative heat load in real time to avoid underestimating the risk of thermal damage, and obtain the current heat load margin accordingly.

[0073] Subsequently, a sterilization risk coefficient is generated based on the device's operating status. This coefficient is composed of multiple observable states, including the stability of the spray medium temperature, the fluctuation of the delivery speed, whether the spray supply is continuous and reliable, and the health of the measurement and data links. Each state quantity is mapped to a risk contribution according to preset rules, and frequent changes caused by short-term disturbances are suppressed through hold or hysteresis mechanisms, ultimately forming a comprehensive coefficient that reflects the current sterilization uncertainty.

[0074] After calculating the upper limit of heat load and the sterilization risk factor, the target PU value corresponding to the product sterilization requirements is used as the benchmark target, and a dynamic target PU value is generated under the following constraints: First, the target PU value shall not be lower than the minimum level of sterilization requirements; Second, the adjustment of the target PU value shall not cause the cumulative heat load to exceed the upper limit of heat load. Specifically, when the sterilization risk factor increases, the operating conditions become unstable, or the uncertainty increases, the target PU value is adjusted conservatively according to the rules; when the device operates stably and the heat load margin is close to the upper limit, the target PU value is suppressed from being adjusted upward or allowed to fall back in order to reduce unnecessary heat input; The adjustment range is determined by the degree of risk and the heat load margin, and a change rate limit is set to prevent the target value from changing drastically.

[0075] For example: The product's thermal sensitivity parameter is mapped to a maximum allowable cumulative heat load upper limit of 100 units (illustrated value). During operation, the cumulative heat load is calculated using the upper boundary of the internal cold point temperature trajectory. The current cumulative value is 82 units, so the heat load margin is 18 units. At the same time, the device's operating status shows that the spray medium temperature is stable, the conveying speed fluctuation is small, the measurement link is healthy, and the sterilization risk factor is low. At this time, the dynamic target PU value can be appropriately adjusted to the lower side of the benchmark target or no longer adjusted upward, provided that it does not fall below the sterilization requirement, in order to reduce unnecessary heat treatment. If the conveying speed subsequently fluctuates frequently and is accompanied by a short-term sudden change in the spray medium temperature, the risk factor increases, leading to an increase in the sterilization risk factor. Even if the heat load margin still exists, the dynamic target PU value will be adjusted to a more conservative direction according to the rules. After the operating conditions stabilize and the stability criteria are continuously met, the risk factor falls back, and the target PU value gradually returns according to the change rate limit to avoid "fluctuating" control oscillations.

[0076] In this way, the target generation module incorporates both the upper limit of heat treatment that the product can withstand and the sterilization uncertainty brought about by equipment operation into the generation logic of the target PU value, so that the target PU value can be adjusted on a basis according to the changes in operating conditions, effectively avoiding excessive heat treatment while ensuring sterilization safety.

[0077] In this embodiment, the joint adjustment module is used to generate joint adjustment instructions. These instructions are calculated online by the rolling optimization control algorithm in each control cycle and are used to coordinate the adjustment of the spray medium temperature and the conveying speed. "Joint" means that under the same prediction and constraint framework, different adjustment methods are compared and selected. Under the premise of meeting the sterilization requirements, the control method with less impact on the product heat load is given priority.

[0078] Among them, the rolling optimization control algorithm refers to a set of control strategies with clear structural characteristics and operation mode. That is, in each control cycle, based on the current measurement and estimation state, it predicts the change trend of PU in the future period, calculates the most suitable control action, executes only one step, and re-predicts and optimizes in the next control cycle. The control cycle is the time scale for repeated operation of the algorithm; the future time step is a discrete representation of the prediction window, used to describe the evolution process of the future PU.

[0079] At the beginning of each control cycle, the real-time PU value, target PU value, current spray medium temperature, conveying speed, and related constraints are updated. The real-time PU value is calculated by the conservative temperature trajectory of the PU calculation module, and the target PU value comes from the dynamic target setting of the target generation module. The deviation between the two constitutes the core constraint basis of rolling optimization.

[0080] During the prediction phase, the algorithm extrapolates the future PU change path based on candidate control actions: the adjustment of the spray medium temperature mainly changes the heat exchange intensity per unit time and has a relatively fast response; the adjustment of the conveying speed affects the cumulative heat input in the future period by changing the distribution of the product's residence time in each processing section. The controller uses the prediction model of the internal cold point temperature trajectory to convert the above effects into temperature changes at each time step in the future, and accumulates them according to the rules consistent with real-time calculation to obtain the future PU trend.

[0081] During the optimization solution, the deviation between the real-time PU value and the target PU value is used as a constraint condition. The focus is on ensuring that the PU value within the prediction window will not be lower than the target requirement, while suppressing excessive heat treatment caused by the PU value being higher than the target value for a long time. The constraints also include limits on the range and rate of change of the spray medium temperature and the conveying speed to ensure the stability and feasibility of the adjustment process.

[0082] Under the premise of satisfying the PU constraint, the algorithm further compares the impact of different control actions on the product's heat load. Specifically, it predicts the cumulative heat load change caused by adjusting only the spray medium temperature, adjusting only the conveying speed, and the combined adjustment when necessary, within the same window. It eliminates the schemes that cannot meet the PU constraint, and then prioritizes the control actions with smaller heat load increments among the feasible schemes. Thus, "prioritizing the adjustment method with less impact on the product's heat load" is implemented as a clear calculation and screening rule, rather than an empirical judgment.

[0083] For example: when the prediction shows that the PU is slightly lower than the target and the device is operating stably, the algorithm may prioritize making up for the PU shortage by fine-tuning the delivery speed to avoid the additional heat load caused by raising the spray temperature; while when the PU is significantly insufficient or the disturbance is strong, it is allowed to prioritize the use of spray medium temperature regulation to ensure that the PU quickly returns to the safe zone, and then gradually reduce unnecessary heat input through rolling optimization in subsequent cycles.

[0084] Through the above method, the joint adjustment module realizes the closed-loop logic of "prediction-constraint-trade-off" in each control cycle: first, it predicts the future changes of PU, then it forms constraints based on the deviation between the real-time PU and the target PU, and finally, under the premise of meeting the sterilization requirements, it automatically selects the control action that has less impact on the product's heat load, thereby achieving a coordinated unity between sterilization safety and heat treatment restraint.

[0085] In this implementation, when a change in container identification information is detected, a product switching process is executed, including: recording the thermal state of each spray zone at the moment of switching; predicting the initial thermal environment when the first batch of new product containers enter each spray zone; and, based on the prediction, making advance correction to the spray medium temperature setpoint to ensure a continuous and smooth PU accumulation process for the first batch of new product containers.

[0086] After the batch is completed, the cold spot temperature trajectory recorded within the batch and the corresponding PU calculation results are statistically analyzed. When a systematic trend of PU deviation is detected in consecutive batches, the model parameters related to spray heat exchange in the heat transfer state model are automatically corrected to reduce the PU prediction error in subsequent batches and improve the accuracy of online monitoring.

[0087] A systematic trend can be determined when one or a combination of the following characteristics are met:

[0088] The PU deviation is consistently positive or negative across multiple consecutive batches, and the direction is consistent.

[0089] Although the PU deviation fluctuates slightly, its mean or median value changes monotonically with the batch number, for example, increasing or decreasing batch by batch.

[0090] The distribution center of the PU bias is significantly off from zero and exceeds the range of random noise in a statistical sense, rather than fluctuating symmetrically around zero.

[0091] This trend usually reflects a slow but continuous change in heat transfer conditions, such as a decrease in spray heat exchange efficiency, changes in nozzle status, or changes in the characteristics of the heat exchange medium, rather than a single disturbance in operating conditions.

[0092] Example 2:

[0093] like Figure 2 As shown, this embodiment further improves upon the design of Embodiment 1. The difference lies in the fact that, in the actual operation of Embodiment 1, it was found that under continuous multi-batch operation conditions, even if the real-time PU value within a single batch could stably follow the target PU value, the PU statistical results between batches might still gradually deviate from the predetermined level. Further analysis indicated that this deviation was not caused by instantaneous operating condition fluctuations, but rather by the slow change in spray heat exchange conditions over operating time, leading to a gradual inaccuracy in the heat transfer state model's characterization of the cold point temperature response. This resulted in a failure to maintain long-term consistency between PU prediction and the actual process across batches. Based on this, this embodiment provides an online monitoring and intelligent adjustment system for the PU value of a wine sterilization process, which further includes:

[0094] Model validation module: After the batch ends, a model consistency validation and parameter update triggering mechanism for cross-batch operation is introduced to determine whether the current heat transfer state model is still applicable to subsequent batch operation, and to adaptively correct the model when the update conditions are met, and to keep the model parameters frozen when the conditions are not met.

[0095] A heat transfer state model is a model used to describe the heat exchange behavior between the spray medium and the product, and to infer the internal cold spot temperature trajectory. Its parameters usually reflect process characteristics such as spray heat transfer intensity and response hysteresis characteristics. Model consistency verification refers to comparing the performance of the model calculation results on a batch scale with comparable PU statistical characteristics to determine whether the model can still stably reflect the actual heat transfer state. The parameter update trigger mechanism is used to determine whether the model parameters can be corrected, while model parameter freezing means that the parameters are explicitly prohibited from changing with short-term data fluctuations when the trigger conditions are not met.

[0096] In terms of specific implementation, after the batch is completed, the model verification module first performs batch-level summary processing on the cold point temperature trajectory recorded in the batch and its corresponding PU calculation results. This summary extracts statistical quantities that can represent the overall heat transfer and sterilization characteristics of the batch, such as the average level of the cold point temperature trajectory in the key section, the overall deviation of the PU cumulative result relative to the target PU value, etc. In this way, high-frequency operating data is converted into low-frequency features suitable for cross-batch comparison, reducing the impact of accidental disturbances on the judgment results.

[0097] Subsequently, the model validation module compares the statistical characteristics of the current batch with the reference characteristics formed under the same or comparable process conditions in historical batches. The comparable process conditions here include at least the premise that the product type is consistent, the target PU value range is consistent, and the equipment operation mode has not undergone structural changes. The focus of the comparison is not the magnitude of a single deviation, but the evolution characteristics of the deviation in the batch sequence, that is, whether it shows a consistent direction, a continuous existence, or a gradually amplifying trend.

[0098] When the statistical results show that the PU-related deviation fluctuates randomly around zero in multiple batches, and the fluctuation amplitude is consistent with previous operating experience, the model verification module determines that the model is still applicable under the current heat transfer state, and at this time the model parameters are kept frozen. Freezing does not mean stopping the use of the model, but rather prohibiting its parameters from being directly modified by online or batch-level data in subsequent calculations, thereby ensuring that the model structure on which the real-time control link depends remains stable.

[0099] Conversely, when the model verification module detects that the batch-level PU deviation exhibits stable and repeatable offset characteristics, such as consistently being higher or lower in several consecutive batches, and this phenomenon cannot be explained by the abnormal operation status of a single batch, the module determines that there is a systematic mismatch between the heat transfer state model and the actual spray heat exchange conditions. In this case, the model verification module releases the parameter update permission, triggering the subsequent adaptive correction process of model parameters related to spray heat exchange, so that the model can be refitted to the current equipment and operating conditions.

[0100] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

[0101] The above description is merely a preferred embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present application, based on the technical solution and concept of the present application, should be covered within the scope of protection of the present application.

Claims

1. A system for online monitoring and intelligent adjustment of PU value in the sterilization process of alcoholic beverages, characterized in that, The system includes: Identification and loading module: Before the container to be sterilized enters the sterilization device, it reads the container identification information and loads the product thermal parameter set corresponding to the container accordingly; Multi-source acquisition module: During the sterilization process, the temperature of the spray medium and the conveying speed are collected according to the spray zone, and the temperature of the outer surface of the container is collected simultaneously to form a continuous temperature sequence; Reference temperature measurement module: Selects the accompanying reference container and obtains its internal temperature sequence, which is used to construct internal temperature calibration constraints; Heat transfer modeling module: Based on the spray medium temperature, conveying speed and the product's thermosensitive parameter set, establish a heat transfer state model that can describe the internal candidate cold points; Fusion estimation module: Input the continuous temperature sequence of the outer surface of the container and the internal temperature sequence of the reference container into the fusion estimation algorithm to correct the heat transfer state model online and obtain the internal cold point temperature trajectory and its uncertainty range; Cold spot reconstruction module: By comparing the deviation between the internal temperature sequence of the reference container and the prediction results of each candidate cold spot, the internal cold spot location is reconstructed in real time and the heat transfer state model is updated. PU calculation module: Based on the uncertainty range of the cold spot temperature trajectory, a conservative temperature trajectory that is not conducive to sterilization is selected and time integration is performed to obtain the real-time PU value; Target generation module: Generates dynamic target PU values ​​based on the product's thermal parameter set and the device's operational risk status; Joint adjustment module: compares the real-time PU value with the target PU value, generates and executes joint adjustment commands for the spray medium temperature and conveying speed.

2. The online monitoring and intelligent adjustment system for PU value in the sterilization process of alcoholic beverages according to claim 1, characterized in that, The fusion estimation algorithm includes: performing first-order dynamic filtering on the temperature sequence of the outer surface of the container to obtain a denoised temperature response; performing time alignment processing on the temperature sequence of the reference container; and adjusting the heat transfer gain parameters in the heat transfer state model online based on the instantaneous deviation between the predicted internal temperature of the outer surface and the internal temperature of the reference container within each control cycle, so that the predicted internal cold point temperature gradually converges to the reference constraint range.

3. The online monitoring and intelligent adjustment system for PU value in the alcohol sterilization process according to claim 1, characterized in that, The process of reconstructing the location of the internal cold spot includes: In the heat transfer state model, N candidate cold point positions are discretized along the axial and radial directions of the container; the internal temperature prediction trajectory is calculated for each candidate cold point; the cumulative residual between the predicted trajectory and the internal temperature sequence of the reference container is calculated within a time window; the candidate cold point with the smallest cumulative residual is selected as the actual cold point under the current sterilization condition, and this cold point is used for subsequent PU calculations.

4. The online monitoring and intelligent adjustment system for PU value in the sterilization process of alcoholic beverages according to claim 1, characterized in that, The method for selecting the conservative temperature trajectory includes: Within the uncertainty range corresponding to the internal cold point temperature trajectory, the lower limit of the temperature is selected as the input for calculating the sterilization effect; and when the conveying speed changes or the spray medium temperature changes abruptly, the uncertainty range is expanded to improve the safety margin of the real-time PU value under dynamic operating conditions.

5. The online monitoring and intelligent adjustment system for PU value in the sterilization process of alcoholic beverages according to claim 1, characterized in that, The method for generating the dynamic target PU value includes: The product's thermal parameters are mapped to the maximum allowable cumulative heat load; the device's operating status is mapped to the sterilization risk coefficient; the adjustment direction and magnitude of the target PU value are determined jointly based on the upper limit of the heat load and the sterilization risk coefficient, so that the target PU value meets the sterilization requirements while avoiding unnecessary heat treatment.

6. The online monitoring and intelligent adjustment system for PU value in the sterilization process of alcoholic beverages according to claim 1, characterized in that, The joint adjustment command is generated by the rolling optimization control algorithm. In each control cycle, the rolling optimization control algorithm: predicts the PU change trend in the future time step; uses the deviation between the real-time PU value and the target PU value as a constraint; and, under the premise of satisfying the PU constraint, prioritizes the control action with less impact on the product heat load, either the spray medium temperature adjustment or the conveying speed adjustment.

7. The online monitoring and intelligent adjustment system for PU value in the sterilization process of alcoholic beverages according to claim 1, characterized in that, When a change in container identification information is detected, the product switching process is executed, including: recording the thermal state of each spray zone at the moment of switching; predicting the initial thermal environment when the first batch of new product containers enter each spray zone; and, based on the prediction, making advance correction to the spray medium temperature setpoint to ensure that the PU accumulation process of the first batch of new product containers is continuous and smooth.

8. The online monitoring and intelligent adjustment system for PU value in the sterilization process of alcoholic beverages according to claim 1, characterized in that, After the batch is completed, the cold spot temperature trajectory recorded in the batch and the corresponding PU calculation results are statistically analyzed. When a systematic trend of PU deviation is detected in consecutive batches, the model parameters related to spray heat exchange in the heat transfer state model are automatically corrected.

9. The online monitoring and intelligent adjustment system for PU value in the sterilization process of alcoholic beverages according to claim 1, characterized in that, Also includes: Model validation module: After the batch ends, a model consistency validation and parameter update triggering mechanism for cross-batch operation is introduced to determine whether the current heat transfer state model is still applicable to subsequent batch operation, and to adaptively correct the model when the update conditions are met, and to keep the model parameters frozen when the conditions are not met.

Images