Intelligent control system for freeze-drying parameters

Abstract

The present application relates to the technical field of data processing, and more particularly to a freeze-drying parameter intelligent regulation system, which comprises a collection module, a determination module, a stage determination module, a type determination module, a control module, a verification module and an adjustment module. The present application constructs a quality index through multi-dimensional quantitative characterization data, and identifies whether the sublimation stage is abnormal by analyzing the time sequence coupling characteristics of ice crystal displacement rate and surface collapse degree. Further, the present application distinguishes the abnormal reasons by using the combination characteristics of rate offset and collapse change rate, so as to take targeted strategies. The regulation effect is verified by the rehydration qualification rate. Based on the historical abnormal density, the threshold is optimized, and the precise positioning and differential treatment of abnormal root causes are realized through multi-parameter coupling analysis, thereby effectively solving the problems of low abnormal recognition accuracy and low freeze-drying control precision caused by excessive dependence on static threshold and inability to adapt to complex working conditions.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to an intelligent control system for freeze-drying parameters. Background Technology

[0002] With the upgrading of health consumption and the popularization of freeze-drying technology, high-pectin fruits such as apples, peaches, apricots, and hawthorns have become important categories in the freeze-dried food market due to their rich dietary fiber and unique taste. However, the cell walls of high-pectin fruits are rich in pectin polysaccharides, which exhibit high hydrophilicity, easy gelation, and low glass transition temperature during freeze-drying. This makes these materials prone to quality problems such as surface crusting, internal collapse, and color deterioration during the sublimation drying stage. Traditional freeze-drying equipment relies on fixed process parameters and cannot sense the real-time changes in the material's state during the drying process. It is difficult to cope with the complex nonlinear behavior of high-pectin fruits, from differences in raw material maturity to the dynamic evolution of the drying process. This results in high product collapse rates, poor rehydration, and high energy consumption, which seriously restricts the automation level and the improvement of product added value in the freeze-drying processing of high-pectin fruits.

[0003] Current industrial freeze-drying equipment primarily employs a temperature-pressure dual-parameter feedback control mode. By arranging thermocouples and vacuum sensors within the freeze-drying chamber, it monitors the plate temperature, material temperature, and chamber pressure in real time, comparing these parameters with a preset freeze-drying curve. A PLC or PID controller adjusts the heating power and vacuum pump operation accordingly. For different materials, the equipment typically has a built-in offline process library, allowing operators to manually select the corresponding freeze-drying curve (pre-freezing temperature, sublimation temperature, desorption temperature, etc.) based on the material type. Some high-end equipment is beginning to incorporate machine vision systems to identify material types during the feeding stage and automatically match them to the process library, reducing errors from manual parameter selection. In continuous freeze-drying production lines, conveyor speed, as a macroscopic process parameter, is typically set empirically based on the material type and thickness and remains constant throughout the production process.

[0004] Therefore, existing technologies suffer from the following problems: First, they have a single sensing dimension, failing to capture the essential changes in material state. Temperature and pressure parameters reflect the state of the cavity environment, and there is an indirect and non-linear mapping relationship with the core quality indicators of the material itself. When internal collapse or surface crusting occurs, temperature and pressure signals often lag or show no significant change, causing the system to fail to respond in a timely manner. Second, anomaly localization is ambiguous, making it difficult to distinguish between process problems with different mechanisms. When the sublimation rate decreases, existing systems can only determine that the "temperature deviates from the curve," but cannot distinguish whether this anomaly is due to insufficient heating or surface crusting, let alone identify whether structural collapse has occurred. This control logic, which knows what but not why, leads to a single control method, making it prone to misadjustment or underadjustment. Third, they lack adaptive optimization capabilities and cannot cope with material fluctuations. Once the offline process library is set, it remains fixed and cannot be dynamically optimized according to actual production conditions such as differences in raw material maturity, batch moisture content fluctuations, and changes in environmental temperature and humidity. When anomalies occur frequently or control fails, the system cannot autonomously adjust detection thresholds or control strategies, relying heavily on manual intervention by engineers, making it difficult to achieve stable and efficient continuous production. Summary of the Invention

[0005] To address this, the present invention provides an intelligent control system for freeze-drying parameters, which overcomes the problems of low accuracy in anomaly identification and low precision in freeze-drying control caused by over-reliance on static thresholds and inability to adapt to complex working conditions in the prior art through multi-dimensional data analysis and dynamic adjustment mechanisms.

[0006] To achieve the above objectives, the present invention provides an intelligent control system for freeze-drying parameters, comprising:

[0007] The data acquisition module is used to collect data in real time on each material to be tested on the freeze-drying production line, including the ice crystal migration rate and surface collapse degree in the sublimation drying zone, the color difference change rate in the desorption drying zone, the rehydration rate in the discharge zone, and the rehydration qualification rate.

[0008] The determination module is used to determine the occurrence of a freeze-drying abnormality event based on the comparison result between the quality index and the preset quality threshold. The quality index is calculated by weighting the surface collapse degree, the color difference change rate and the rehydration rate.

[0009] A stage determination module is used to determine the occurrence of sublimation drying anomaly based on the freeze-drying anomaly event and according to the numerical characteristics of the changing coupling degree, wherein the changing coupling degree is determined based on the temporal variation characteristics of the ice crystal migration rate and the surface collapse degree.

[0010] The type determination module is used to determine the anomaly type as heat transfer anomaly or structural collapse anomaly based on the sublimation drying anomaly and according to the temporal correlation characteristics of the ice crystal migration rate and the surface collapse degree.

[0011] The control module is used to adjust the heating power of the sublimation drying zone according to the ice crystal moving rate based on the heat transfer anomaly type, and to adjust the conveying speed and heating power according to the surface collapse degree based on the structural collapse type.

[0012] The verification module is used to perform freeze-drying verification based on the adjusted heating power and the conveying speed, according to the change characteristics of the rehydration qualification rate within a preset verification time, and to issue an early warning based on the verification results.

[0013] The adjustment module is used to adjust the preset quality threshold according to the temporal characteristics of the freeze-drying abnormal events occurring within a preset observation period.

[0014] Furthermore, when the quality index is less than the preset quality threshold, a freeze-drying abnormality event is determined to have occurred.

[0015] Furthermore, the quality index is calculated by weighting the surface collapse degree, the color difference change rate, the rehydration rate, the preset collapse weight, the preset color difference weight, and the preset rehydration weight.

[0016] Furthermore, when the changed coupling degree is greater than a preset coupling degree threshold, it is determined that a sublimation drying anomaly has occurred.

[0017] Furthermore, the variable coupling degree is calculated based on the directional synchronization factor, nonlinear deviation degree, preset directional weight, and preset nonlinear weight. The directional synchronization factor is calculated based on the cosine similarity between the ice crystal moving rate and the surface collapse degree, and the nonlinear deviation degree is calculated based on the Pearson correlation coefficient between the ice crystal moving rate and the surface collapse degree.

[0018] Furthermore, the type determination module includes:

[0019] The feature calculation unit is used to calculate the rate offset based on all the ice crystal pushing rates and the preset target rate within a preset time period, and to calculate the collapse change rate based on all the surface collapse degrees within a preset time period.

[0020] A type determination unit, connected to the feature calculation unit, is used to determine the anomaly type as the heat transfer anomaly type when the rate offset is less than a first preset threshold and the collapse change rate is less than or equal to a second preset threshold, and to determine the anomaly type as the structural collapse type when the rate offset is less than the first preset threshold and the collapse change rate is greater than the second preset threshold.

[0021] Furthermore, the control module includes:

[0022] The first control unit is used to calculate the rate offset based on the heat transfer anomaly type, the ice crystal pushing rate and the preset target rate, and adjust the heating power based on the comparison result of the rate offset and the preset offset threshold.

[0023] The second control unit is used to adjust the conveying speed and the heating power based on the surface collapse degree, according to the structure collapse type.

[0024] Furthermore, the second control unit includes:

[0025] A collapse deviation calculation subunit is used to calculate the collapse deviation based on the surface collapse degree and a preset collapse threshold.

[0026] A speed adjustment subunit, which is connected to the collapse deviation calculation subunit, is used to adjust the transmission speed according to the comparison result of the collapse deviation and the preset collapse deviation threshold and the preset speed adjustment coefficient.

[0027] A power adjustment subunit, which is connected to the collapse deviation calculation subunit, is used to adjust the heating power according to the comparison result of the collapse deviation and the preset collapse deviation threshold and the preset power adjustment coefficient.

[0028] Furthermore, the verification module includes:

[0029] A change calculation unit is used to calculate the change rate of the pass rate based on all the rehydration pass rates within the preset verification time.

[0030] A fluctuation calculation unit is used to calculate the fluctuation value of the pass rate based on the pass rate of all the rehydration pass rates within the preset verification period.

[0031] A verification unit, which is connected to the change calculation unit and the fluctuation calculation unit respectively, is used to issue an early warning when the pass rate change rate is less than a preset pass change threshold and the pass rate fluctuation value is greater than a preset pass fluctuation threshold.

[0032] Furthermore, the adjustment module includes:

[0033] An anomaly density calculation unit is used to calculate the anomaly density based on the timestamps of all freeze-drying anomalies within the preset observation period.

[0034] An adjustment unit, connected to the abnormal density calculation unit, is used to adjust the preset quality threshold based on the comparison result between the abnormal density and the preset abnormal density threshold.

[0035] Compared with existing technologies, the beneficial effects of this invention are as follows: by using parameters such as ice crystal migration rate, surface collapse degree, color difference change rate, and rehydration rate, sublimation mass transfer efficiency, structural stability, thermal damage degree, and final performance are directly mapped, forming a multi-dimensional quantitative characterization from process to quality; by constructing a quality index, real-time evaluation of the overall product status is achieved; by analyzing the temporal coupling characteristics of ice crystal migration rate and surface collapse degree, the abnormality can be accurately identified as originating from the sublimation stage, because there is a causal chain of "heat and mass transfer obstruction - structural collapse" in their physical mechanisms; further, the rate bias of the two is utilized... The combined characteristics of displacement and collapse change rate can distinguish between anomalies caused by insufficient heating or structural fragility, thereby guiding the control module to adopt targeted strategies; the change in rehydration qualification rate verifies the control effect, forming a closed loop; based on historical anomaly density, the detection threshold is adaptively optimized, transforming the physicochemical mechanism of the freeze-drying process into calculable data relationships, and through multi-parameter coupling analysis, the root cause of anomalies can be accurately located and differentiatedly handled, ultimately achieving the goals of improving quality, reducing consumption, and increasing efficiency. This effectively solves the problems of low anomaly identification accuracy and low freeze-drying control precision caused by over-reliance on static thresholds and inability to adapt to complex working conditions.

[0036] Furthermore, by mapping multi-source process parameters to a unified quality scale and achieving objective quantitative identification of anomalies through threshold comparison, the quality index is a weighted fusion of surface collapse degree, color difference change rate, and rehydration rate. Surface collapse degree directly reflects the structural integrity of the material; the larger the value, the worse the quality. Color difference change rate characterizes the degree of thermal damage; the faster the change, the worse the quality. Rehydration rate reflects the final performance; the faster the rate, the better the quality. Its value is negatively correlated with the overall product quality; that is, the smaller the index, the better the quality. Therefore, when the real-time calculated quality index is lower than the preset threshold, it means that the structure, color, or rehydration performance of the material has deviated from the qualified standard. The system determines that an abnormal event has occurred based on this and reverses the final product quality requirements into process parameter constraints. Through real-time monitoring and threshold comparison, a rapid response to anomalies in the freeze-drying process is achieved, providing a clear trigger basis for subsequent precise positioning and control.

[0037] Furthermore, by weighted and fused three parameters—surface collapse degree, color difference change rate, and rehydration rate—the quality index is calculated, deconstructing the core quality dimensions of freeze-dried products into quantifiable process characterization indicators. Multi-objective comprehensive evaluation is achieved through weight allocation. Surface collapse degree directly corresponds to the integrity of the material structure, reflecting whether collapse and deformation occur during freeze-drying; color difference change rate characterizes color stability, quantifying the degree of thermal damage or enzymatic browning; and rehydration rate reflects functional recovery, directly measuring the product's final application value. These three parameters define product quality from three independent dimensions: "shape, color, and function," complementing each other and being irreplaceable. By introducing weights, the system can dynamically adjust the contribution of each dimension according to different product positioning, achieving configurable quality evaluation standards. This transforms multi-dimensional quality objectives into a single comprehensive indicator, preserving the physical meaning of each parameter while allowing flexible control of evaluation focus through weight allocation, providing a comprehensive and targeted quantitative basis for subsequent anomaly detection.

[0038] Furthermore, by transforming the physical coupling relationship of "heat and mass transfer-structural response" in the sublimation stage into a quantifiable data correlation index, an inherent causal chain exists between the ice crystal migration rate and surface collapse degree during freeze-drying sublimation. When heating is insufficient or mass transfer is hindered, the ice crystal migration rate decreases. If this continues for too long or there is localized overheating, it may lead to structural weakening, resulting in an increase in surface collapse degree. Conversely, if the material itself has a fragile structure, even if the ice crystal migration rate is normal, abnormal surface collapse degree may still occur. Therefore, if a strong coupling characteristic appears between the ice crystal migration rate and surface collapse degree, such as a highly consistent direction of change or a significant linear correlation, it often indicates that the sublimation process has deviated from the normal state. The degree of change coupling is a comprehensive index that quantifies the dynamic relationship between the two by integrating their directional synchronicity and linear correlation strength. When this index exceeds a preset threshold, it means that an abnormal synergistic change has occurred between the ice crystal migration rate and surface collapse degree, thus determining that an anomaly has occurred in the sublimation stage. By capturing the coupling abrupt change between parameters, early warning and precise location of anomalies in the sublimation stage are achieved, providing a reliable basis for subsequent differentiation of anomaly types.

[0039] Furthermore, by defining the degree of change coupling as a weighted fusion of the directional synchronization factor and the nonlinear deviation, the dynamic coupling strength between ice crystal propagation rate and surface collapse degree is comprehensively quantified from two orthogonal dimensions: "direction of change" and "linear relationship." The directional synchronization factor, calculated based on cosine similarity, captures the consistency between ice crystal propagation rate and surface collapse degree in the direction of change. When both rise or fall synchronously, it indicates a strong correlation between the sublimation process and the structural response. The nonlinear deviation is calculated based on the Pearson correlation coefficient, and its complement (1-|ρ|) reflects the degree of lack of a linear relationship between ice crystal propagation rate and surface collapse degree. When the two exhibit complex nonlinear coupling, such as collapse causing a sharp drop in ice crystal propagation rate while the surface collapse degree slowly increases, this index increases significantly. By flexibly adjusting the sensitivity to "directional synchronization" and "nonlinear coupling" through weight allocation, a comprehensive index is constructed to fully characterize the complex relationship between ice crystal migration rate and surface collapse. This transforms the multidimensional coupling characteristics of "heat transfer and mass transfer-structural response" in the physical process into a calculable data relationship. By integrating directional and correlation information, early and accurate identification of anomalies in the sublimation stage is achieved, providing richer decision-making basis for distinguishing between "heat transfer anomalies" and "structural collapse".

[0040] Furthermore, by using a combined threshold of rate offset and collapse change rate to determine the anomaly type, the physical causal chain of "energy supply-structural response" in the sublimation stage is deconstructed into quantifiable data feature pairs. Among them, the rate offset directly characterizes whether the sublimation driving force is sufficient. When it is less than the first preset threshold, it indicates that the ice crystal sublimation rate is significantly lower than expected, which may be due to insufficient heating or obstruction of water vapor escape. The collapse change rate quantifies whether the structural stability is lost. When it exceeds the second preset threshold, it indicates that the material surface is undergoing continuous collapse. The combination of the two constitutes a unique fingerprint of the anomaly type: if the rate offset is negatively out of tolerance while the collapse change rate is stable, it indicates that sublimation is blocked but the structure is still intact, which is a typical heat transfer anomaly, that is, insufficient heating or poor mass transfer but no damage has been caused; if the rate offset is negatively out of tolerance and the collapse change rate increases significantly, it indicates that the obstruction of sublimation has led to continuous structural deterioration, which is a structural collapse type, that is, insufficient heat resistance of the material or pre-freezing defects. By combining the two orthogonal dimensions of "whether there is a lack of energy" and "whether there is damage", the cause-and-effect tracing of complex physical processes is realized through simple threshold comparison, providing a clear and reliable decision basis for subsequent differentiated control.

[0041] Furthermore, a divide-and-conquer control strategy is adopted to increase power for abnormal heat transfer and reduce speed and temperature for structural collapse, thereby achieving precise intervention for abnormal types. A causal closed loop is established to map the root cause of the abnormality to the control measures. For the abnormal heat transfer type where sublimation is hindered but the structure is still stable, the comparison between the rate deviation and the preset deviation threshold is used: since the rate is low at this time, when the absolute value of the rate deviation is greater than the preset deviation threshold, it indicates that the sublimation driving force is seriously insufficient, and the heating power needs to be increased to enhance the heat input, thereby increasing the ice crystal sublimation rate. The adjustment range is calculated based on the product of the relative deviation and the preset first adjustment coefficient, realizing proportional control where the larger the deviation, the stronger the compensation. For structural collapse caused by hindered sublimation and structural deterioration, since the material has already suffered structural damage, simply increasing the heat will exacerbate the collapse. Therefore, it is necessary to reduce the heating power to weaken the thermal shock and at the same time reduce the conveying speed to extend the residence time in the pre-freezing zone, thereby strengthening the material skeleton from the root. The two control modes of energy supply and structural protection are decoupled, and the control strategy is automatically switched by identifying the abnormal type. This avoids under-adjustment when there is abnormal heat transfer and over-adjustment when there is structural collapse, thus achieving adaptive optimization control of the freeze-drying process.

[0042] Furthermore, a graded response mechanism for collapse deviation is used to achieve coordinated control of structural collapse anomalies. The degree of surface collapse deviation is used as the trigger for control intensity. Collapse deterioration is prevented through a dual approach of deceleration to strengthen the structure and cooling to reduce heat load. The real-time surface collapse degree is compared with a preset collapse threshold to calculate the collapse deviation, which directly quantifies the degree of structural damage exceeding the limit. When the collapse deviation exceeds the preset threshold, it indicates that the collapse has exceeded the allowable range and immediate intervention is required. This involves reducing the conveying speed to extend the material's residence time in the pre-freezing zone, promoting ice crystal refinement and skeleton strengthening, thereby fundamentally improving the structure's heat resistance. It also involves reducing heating power to decrease heat input to the sublimation zone, preventing high temperatures from exacerbating collapse. A proportional control mechanism, where the larger the deviation, the stronger the deceleration and cooling, transforms the degree of structural damage into a linear mapping of deceleration and cooling intensity. Through dual-parameter coordinated adjustment, the further development of the current collapse is curbed, and the material's anti-collapse ability is fundamentally improved, achieving a comprehensive solution for structural collapse anomalies.

[0043] Furthermore, an objective evaluation of the control effect is achieved through a dual verification mechanism. The freeze-drying process verification is deconstructed into two orthogonal dimensions: trend effectiveness and stability reliability. The combined characteristics of these two dimensions are used to determine whether the control truly solves the problem. The slope of the linear change in the rehydration pass rate within the verification period is calculated to quantify the improvement trend of quality after control. If the change rate of the pass rate is less than the preset pass rate change threshold, it indicates that the control has failed to effectively improve the rehydration performance of the product. The standard deviation of the rehydration pass rate is calculated to characterize the consistency level of quality. If the fluctuation value of the pass rate is greater than the preset fluctuation threshold, it indicates that the product quality is unstable and the control effect is unreliable. An early warning is issued when both exceed the standard. When the change rate of the pass rate is less than the threshold and the fluctuation value is greater than the threshold, it means that the control has neither improved the quality nor reduced the product consistency. This is a typical control failure mode, and manual intervention should be notified immediately. By combining the two independent evaluation criteria of whether the quality has improved and whether the quality is stable, the dual condition constraints effectively avoid the misjudgment that may be caused by a single indicator, and achieve a comprehensive and objective evaluation of the control effect. This provides a reliable decision-making basis for subsequent process library correction and threshold adaptive adjustment.

[0044] Furthermore, a quality threshold adaptive adjustment mechanism driven by anomaly density is used to dynamically optimize detection sensitivity. The underlying logic involves using the frequency characteristics of historical anomalies as feedback signals for threshold optimization, and correcting the judgment criteria by quantifying the deviation between the anomaly density and the preset threshold. The timestamp density of anomalies within a preset observation period characterizes the current anomaly triggering frequency of the system. When the anomaly density exceeds the preset anomaly density threshold, it indicates that the current quality threshold is too strict, leading to frequent anomaly triggering. In this case, the quality threshold is increased to reduce system sensitivity and unnecessary intervention. The density of historical anomalies is used as a negative feedback signal for threshold optimization. By dynamically balancing detection sensitivity and system stability, the quality threshold is always kept at the optimal operating point. This avoids frequent false alarms that interfere with production due to excessively low thresholds, while also preventing missed anomalies due to excessively high thresholds, ultimately achieving adaptive optimization and long-term stable operation of anomaly detection performance. Attached Figure Description

[0045] Figure 1 This is a schematic diagram of the intelligent control system for freeze-drying parameters in this embodiment;

[0046] Figure 2 This is a logic diagram for determining the occurrence of a freeze-drying abnormal event in this embodiment;

[0047] Figure 3 This is the logic diagram for determining whether a sublimation drying abnormality has occurred in this embodiment;

[0048] Figure 4 This is a logic diagram for determining the exception type in the type determination module of this embodiment. Detailed Implementation

[0049] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0050] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0051] Please see Figure 1 As shown, this is a schematic diagram of the intelligent control system for freeze-drying parameters in this embodiment. This embodiment provides an intelligent control system for freeze-drying parameters, including:

[0052] The data acquisition module is used to collect data in real time on each material to be tested on the freeze-drying production line, including the ice crystal migration rate and surface collapse degree in the sublimation drying zone, the color difference change rate in the desorption drying zone, the rehydration rate in the discharge zone, and the rehydration qualification rate.

[0053] The determination module is used to determine the occurrence of a freeze-drying abnormality event based on the comparison result between the quality index and the preset quality threshold. The quality index is calculated by weighting the surface collapse degree, the color difference change rate and the rehydration rate.

[0054] A stage determination module is used to determine the occurrence of sublimation drying anomaly based on the freeze-drying anomaly event and according to the numerical characteristics of the changing coupling degree, wherein the changing coupling degree is determined based on the temporal variation characteristics of the ice crystal migration rate and the surface collapse degree.

[0055] The type determination module is used to determine the anomaly type as heat transfer anomaly or structural collapse anomaly based on the sublimation drying anomaly and according to the temporal correlation characteristics of the ice crystal migration rate and the surface collapse degree.

[0056] The control module is used to adjust the heating power of the sublimation drying zone according to the ice crystal moving rate based on the heat transfer anomaly type, and to adjust the conveying speed and heating power according to the surface collapse degree based on the structural collapse type.

[0057] The verification module is used to perform freeze-drying verification based on the adjusted heating power and the conveying speed, according to the change characteristics of the rehydration qualification rate within a preset verification time, and to issue an early warning based on the verification results.

[0058] The adjustment module is used to adjust the preset quality threshold according to the temporal characteristics of the freeze-drying abnormal events occurring within a preset observation period.

[0059] In this embodiment, the data acquisition module addresses the characteristics of high-pectin fruits (such as apples, peaches, apricots, and hawthorns) on continuous freeze-drying production lines, which are prone to structural collapse, browning, and decreased rehydration performance. Multiple types of vision and physical property sensors are deployed at key process points to collect core macroscopic parameters characterizing the material's state in real time: In the sublimation drying zone, a machine vision system continuously tracks the movement of the drying interface within the porous framework, calculating the increase in the thickness of the drying layer per unit time to obtain the ice crystal migration rate, which directly reflects the sublimation mass transfer efficiency; simultaneously, three-dimensional point cloud reconstruction technology is used to quantify the changes in the depth and area of ​​surface depressions. The process involves several steps: First, surface collapse is measured to characterize the integrity of the material's structure. Second, in the analytical drying zone, a multispectral camera collects the chromaticity values ​​of the material in the Lab color space, and the change in color difference per unit time is calculated to obtain the color difference rate. This parameter quantifies the degree of color deterioration caused by thermal damage or enzymatic browning. Third, in the discharge zone, a high-speed camera and dynamic weighing system record the time required for the freeze-dried product to recover its mass to a specified ratio in standard temperature water, obtaining the rehydration rate. A destructive sampling test is then conducted to determine the percentage of qualified samples, yielding the rehydration pass rate. Both parameters reflect the final performance of the product. These parameters provide a direct and quantitative data foundation for subsequent anomaly detection from four dimensions: sublimation mass transfer, structural stability, color retention, and functional recovery.

[0060] By using parameters such as ice crystal migration rate, surface collapse degree, color difference change rate, and rehydration rate, sublimation mass transfer efficiency, structural stability, thermal damage degree, and final performance are directly mapped, forming a multi-dimensional quantitative characterization from process to quality. A quality index is constructed to achieve real-time evaluation of the product's overall condition. By analyzing the temporal coupling characteristics of ice crystal migration rate and surface collapse degree, it is possible to accurately identify whether anomalies originate from the sublimation stage, as there is a causal chain of "impaired heat and mass transfer - structural collapse" in their physical mechanisms. Furthermore, the rate offset and collapse change rate of these two parameters are further utilized... By combining features, the system distinguishes between anomalies caused by insufficient heating or structural fragility, thereby guiding the control module to adopt targeted strategies. The system verifies the control effect by measuring changes in the rehydration pass rate, forming a closed loop. Based on historical anomaly density, the system adaptively optimizes the detection threshold, transforming the physicochemical mechanism of the freeze-drying process into calculable data relationships. Through multi-parameter coupling analysis, the system achieves precise localization and differentiated treatment of the root causes of anomalies, ultimately achieving the goals of improving quality, reducing consumption, and increasing efficiency. This effectively solves the problems of low anomaly identification accuracy and low freeze-drying control precision caused by over-reliance on static thresholds and inability to adapt to complex working conditions.

[0061] Please see Figure 2 As shown, it is a logic diagram for determining the occurrence of a freeze-drying abnormal event in this embodiment. In this embodiment, a freeze-drying abnormal event is determined to have occurred when the quality index is less than the preset quality threshold.

[0062] The preset quality threshold is a benchmark value used to determine whether the quality of freeze-dried products is qualified. It depends on the product's quality requirements standards, material type characteristics, and the statistical analysis results of historical production data. It is usually set between 0.2 and 0.5. In this embodiment, it is set to 0.35, which can effectively capture most batches with abnormal quality and avoid frequent false alarms caused by normal fluctuations, thus balancing detection sensitivity and production continuity.

[0063] By mapping multi-source process parameters to a unified quality scale and achieving objective quantitative identification of anomalies through threshold comparison, the quality index is a weighted fusion of surface collapse degree, color difference change rate, and rehydration rate. Surface collapse degree directly reflects the structural integrity of the material; the larger the value, the worse the quality. Color difference change rate characterizes the degree of thermal damage; the faster the change, the worse the quality. Rehydration rate reflects the final performance; the faster the rate, the better the quality. Its value is negatively correlated with the overall product quality; that is, the smaller the index, the better the quality. Therefore, when the real-time calculated quality index is lower than the preset threshold, it means that the structure, color, or rehydration performance of the material has deviated from the qualified standard. The system determines that an abnormal event has occurred based on this and reverses the final product quality requirements into process parameter constraints. Through real-time monitoring and threshold comparison, a rapid response to anomalies in the freeze-drying process is achieved, providing a clear trigger basis for subsequent precise positioning and control.

[0064] Specifically, the quality index is calculated by weighting the surface collapse degree, the color difference change rate, the rehydration rate, the preset collapse weight, the preset color difference weight, and the preset rehydration weight.

[0065] The preset collapse weight, preset color difference weight, and preset rehydration weight are weighting coefficients assigned to surface collapse degree, color difference change rate, and rehydration rate, respectively, in the quality index calculation. Their values ​​depend on the characteristics of the fruit variety. For example, high-pectin fruits are sensitive to collapse and prone to browning during the resolution stage, as well as the requirements for rehydration in the final use of the product. They are usually set between 0.3 and 0.6, 0.2 and 0.4, and 0.2 and 0.5, respectively. In this embodiment, they are set to 0.45, 0.25, and 0.30, respectively. A moderately high collapse weight is assigned to the characteristic of high-pectin fruits being prone to collapse in order to prioritize the protection of structural integrity. A moderate color difference weight is assigned to monitor thermal damage during the resolution stage while avoiding environmental interference. A moderate rehydration weight is assigned to ensure that the final performance meets the standards. The three factors work together to achieve a comprehensive quantitative evaluation from process state to final quality, enabling the quality index to accurately reflect the overall quality of the product and laying a reliable foundation for subsequent anomaly detection and precise control.

[0066] The quality index is calculated by weighting and fusing three parameters: surface collapse degree, color difference change rate, and rehydration rate. This deconstructs the core quality dimensions of freeze-dried products into quantifiable process characterization indicators, and achieves multi-objective comprehensive evaluation through weight allocation. Surface collapse degree directly corresponds to the integrity of the material structure, reflecting whether collapse and deformation occur during freeze-drying; color difference change rate characterizes color stability, quantifying the degree of thermal damage or enzymatic browning; and rehydration rate reflects functional recovery, directly measuring the product's final application value. These three parameters define product quality from three independent dimensions: "shape," "color," and "use," complementing each other and being irreplaceable. By introducing weights, the system can dynamically adjust the contribution of each dimension according to different product positioning, achieving configurable quality evaluation standards. This transforms multi-dimensional quality objectives into a single comprehensive indicator, preserving the physical meaning of each parameter while allowing flexible control of evaluation focus through weight allocation, providing a comprehensive and targeted quantitative basis for subsequent anomaly detection.

[0067] Please see Figure 3 As shown, it is the logic diagram for determining whether a sublimation drying abnormality has occurred in this embodiment. In this embodiment, a sublimation drying abnormality is determined to have occurred when the change in coupling degree is greater than a preset coupling degree threshold.

[0068] The preset coupling threshold is a benchmark value used to determine whether the dynamic coupling strength between the sublimation rate and the surface collapse degree has reached an abnormal level. It depends on the physical characteristics of different fruit varieties and the tolerance of the production process to structural stability. It is usually set between 0.5 and 0.8. In this embodiment, it is set to 0.65, which can effectively identify the abnormally strong coupling relationship between the sublimation rate and the degree of collapse, and avoid misjudgment caused by coupling fluctuations caused by the material's own characteristics or normal process fluctuations, thus ensuring the accuracy of stage positioning.

[0069] By transforming the physical coupling relationship of "heat and mass transfer-structural response" in the sublimation stage into a quantifiable data correlation index, an inherent causal chain exists between the ice crystal migration rate and surface collapse degree during freeze-drying sublimation. When heating is insufficient or mass transfer is hindered, the ice crystal migration rate decreases. If this decrease persists for too long or there is localized overheating, it may lead to structural weakening, resulting in an increase in surface collapse degree. Conversely, if the material itself has a fragile structure, even if the ice crystal migration rate is normal, abnormal surface collapse degree may still occur. Therefore, if a strong coupling characteristic appears between the ice crystal migration rate and surface collapse degree, such as a highly consistent direction of change or a significant linear correlation, it often indicates that the sublimation process has deviated from the normal state. The degree of change coupling is a comprehensive index that quantifies the dynamic relationship between the two by integrating their directional synchronicity and linear correlation strength. When this index exceeds a preset threshold, it means that an abnormal synergistic change has occurred between the ice crystal migration rate and surface collapse degree, thus determining that an anomaly has occurred in the sublimation stage. By capturing abrupt coupling changes between parameters, early warning and precise location of anomalies in the sublimation stage are achieved, providing a reliable basis for subsequent differentiation of anomaly types.

[0070] Specifically, the variable coupling degree is calculated based on the directional synchronization factor, nonlinear deviation degree, preset directional weight, and preset nonlinear weight. The directional synchronization factor is calculated based on the cosine similarity between the ice crystal pushing rate and the surface collapse degree, and the nonlinear deviation degree is calculated based on the Pearson correlation coefficient between the ice crystal pushing rate and the surface collapse degree.

[0071] The formula for calculating the degree of coupling variation is: γ=|cosθ(v,S)|×α+(1-|ρ(v,S)|)×β, where cosθ(v,S) is the directional synchronization factor, i.e., the cosine similarity between the ice crystal pushing rate and the surface collapse degree; α is the preset directional weight; ρ(v,S) is the Pearson correlation coefficient between the ice crystal pushing rate and the surface collapse degree; and β is the preset nonlinear weight.

[0072] The preset directional weight and preset nonlinear weight are weighting coefficients assigned to the directional synchronization factor and nonlinear deviation degree respectively in the calculation of the degree of change coupling. They depend on the inherent relationship between rate and collapse in the sublimation stage of different fruit varieties and the sensitivity requirements of the process to the two types of coupling features. They are usually set between 0.3 and 0.7 respectively and the sum of the two is 1. In this embodiment, the preset directional weight is set to 0.5 and the preset nonlinear weight is set to 0.5. The balanced weight allocation fully considers the contribution of the synchronicity of rate and collapse in the direction of change to anomaly identification, and also takes into account the nonlinear anomaly features revealed by the degree of loss of the linear relationship between the two. This allows the degree of change coupling to comprehensively capture the abnormal coupling relationship of different modes, avoiding missed or wrong judgments due to emphasizing one feature. It is particularly suitable for complex materials such as high pectin fruits, which may exhibit both directional synchronous collapse and nonlinear structural degradation at the same time in the sublimation stage. It provides a comprehensive and balanced basis for evaluating the coupling strength for the stage determination module.

[0073] By defining the degree of change coupling as a weighted fusion of the directional synchronization factor and the nonlinear deviation, the dynamic coupling strength between ice crystal propagation rate and surface collapse degree is comprehensively quantified from two orthogonal dimensions: "direction of change" and "linear relationship". The directional synchronization factor, calculated based on cosine similarity, captures the consistency between ice crystal propagation rate and surface collapse degree in the direction of change. When both rise or fall synchronously, it indicates a strong correlation between the sublimation process and the structural response. The nonlinear deviation is calculated based on the Pearson correlation coefficient, and its complement (1-|ρ|) reflects the degree of lack of a linear relationship between ice crystal propagation rate and surface collapse degree. When the two exhibit complex nonlinear coupling, such as collapse causing a sharp drop in ice crystal propagation rate while the surface collapse degree slowly increases, this index increases significantly. By flexibly adjusting the sensitivity to "directional synchronization" and "nonlinear coupling" through weight allocation, a comprehensive index is constructed to fully characterize the complex relationship between ice crystal migration rate and surface collapse. This transforms the multidimensional coupling characteristics of "heat transfer and mass transfer-structural response" in the physical process into a calculable data relationship. By integrating directional and correlation information, early and accurate identification of anomalies in the sublimation stage is achieved, providing richer decision-making basis for distinguishing between "heat transfer anomalies" and "structural collapse".

[0074] Please see Figure 4 As shown, this is a logic diagram for determining the exception type by the type determination module in this embodiment. In this embodiment, the type determination module includes:

[0075] The feature calculation unit is used to calculate the rate offset based on all the ice crystal pushing rates and the preset target rate within a preset time period, and to calculate the collapse change rate based on all the surface collapse degrees within a preset time period.

[0076] A type determination unit, connected to the feature calculation unit, is used to determine the anomaly type as the heat transfer anomaly type when the rate offset is less than a first preset threshold and the collapse change rate is less than or equal to a second preset threshold, and to determine the anomaly type as the structural collapse type when the rate offset is less than the first preset threshold and the collapse change rate is greater than the second preset threshold.

[0077] The preset target rate is a standard value of ice crystal migration rate that the material should achieve during the sublimation drying stage, obtained through statistical analysis of historical normal production batch data. It serves as a benchmark reference value for judging whether the current sublimation rate is normal. It depends on the historical normal batch data of different fruit varieties, material characteristics, and the standard operating status of the equipment. It is usually set within ±10% of the average sublimation rate of normal batches. In this embodiment, it is set as the statistical average of 50 historical normal production batches. The target rate obtained based on historical data statistics truly reflects the optimal matching state of equipment-material-process, making the calculation of rate deviation have practical process significance and avoiding the problem of the theoretical set value being out of sync with actual production.

[0078] The preset time limit is the length of time used to calculate the rate offset and collapse change rate. It depends on the total residence time of the material in the sublimation drying zone, the conveying speed, and the frequency of data sampling. It is usually set between 5 minutes and 30 minutes. In this embodiment, it is set to 15 minutes, which can collect enough data points to smooth noise and accurately calculate trends, and will not cause abnormal response delays due to excessively long windows, thus ensuring the timeliness and accuracy of type determination.

[0079] The first preset threshold is a threshold value used to determine whether the rate offset is too low. It depends on the target sublimation rate of different fruit varieties and its allowable fluctuation range. It is usually set between -20% and -5% of the preset target rate. In this embodiment, it is set to -15% of the preset target rate, which can effectively distinguish between normal fluctuations and real heat transfer anomalies, avoid frequent misjudgments due to oversensitivity, and ensure that the problem of insufficient heat transfer is captured in time.

[0080] The second preset threshold is a threshold value used to determine whether the rate of collapse change is too high. It depends on the structural stability of different fruit varieties, the allowable degree of collapse, and the product appearance quality requirements. It is usually set between 0.02 per minute and 0.08 per minute. In this embodiment, it is set to 0.05 per minute, which can identify the critical point at which the structure begins to deteriorate. It will not be misjudged as a structural collapse due to small fluctuations, and it can trigger regulation in time in the early stage of accelerated collapse, so as to protect the integrity of the material structure to the greatest extent.

[0081] By using a combined threshold of rate offset and collapse change rate, the anomaly type can be accurately distinguished, and the physical causal chain of "energy supply-structural response" in the sublimation stage can be deconstructed into quantifiable data feature pairs. Among them, the rate offset directly characterizes whether the sublimation driving force is sufficient. When it is less than the first preset threshold, it indicates that the ice crystal sublimation rate is significantly lower than expected, which may be due to insufficient heating or obstruction of water vapor escape. The collapse change rate quantifies whether the structural stability is lost. When it exceeds the second preset threshold, it indicates that the material surface is undergoing continuous collapse. The combination of the two constitutes a unique fingerprint of the anomaly type: if the rate offset is negatively out of tolerance while the collapse change rate is stable, it indicates that sublimation is blocked but the structure is still intact, which is a typical heat transfer anomaly, that is, insufficient heating or poor mass transfer but no damage has been caused; if the rate offset is negatively out of tolerance and the collapse change rate increases significantly, it indicates that the obstruction of sublimation has led to continuous structural deterioration, which is a structural collapse type, that is, insufficient heat resistance of the material or pre-freezing defects. By combining the two orthogonal dimensions of "whether there is a lack of energy" and "whether there is damage", the cause-and-effect tracing of complex physical processes is realized through simple threshold comparison, providing a clear and reliable decision basis for subsequent differentiated control.

[0082] Specifically, the control module includes:

[0083] The first control unit is used to calculate the rate offset based on the heat transfer anomaly type, the ice crystal pushing rate, and the preset target rate. When the rate offset is greater than the preset offset threshold, the heating power is increased based on the relative deviation between the rate offset and the preset offset threshold and the preset first adjustment coefficient. Here, L'=L×(1+a×|Q-Q'| / Q'), L' is the adjusted heating power, L is the heating power, a is the preset first adjustment coefficient, Q is the rate offset, and Q' is the preset offset threshold.

[0084] The second control unit is used to adjust the conveying speed and the heating power based on the surface collapse degree, according to the structure collapse type.

[0085] The preset offset threshold is a threshold value used to determine whether the rate offset has reached the level that requires the initiation of regulation. It depends on the system's tolerance for rate fluctuations, the response characteristics of the actuator, and the need to avoid frequent fine-tuning. It is usually set between 5% and 15% of the preset target rate. In this embodiment, it is set to 10% of the preset target rate, which not only filters out invalid adjustments caused by normal rate fluctuations and avoids frequent actuator actions, but also ensures that regulation is initiated in a timely manner when the rate deviates significantly from the target, thus achieving a balance between control stability and timely response.

[0086] The preset first adjustment coefficient is a proportional coefficient that controls the adjustment range of heating power. It depends on the response capability of the heating system, the thermal sensitivity of the material, and the need to avoid overshoot. It is usually set between 0.2 and 0.8. In this embodiment, it is set to 0.5, which can ensure sufficient adjustment to correct rate deviation, while avoiding the risk of new collapse caused by temperature overshoot due to excessive adjustment. It is particularly suitable for heat-sensitive materials such as high-pectin fruits.

[0087] By employing a divide-and-conquer control strategy that increases power for abnormal heat transfer and reduces speed and temperature for structural collapse, precise intervention for abnormal types is achieved. A causal closed loop is established, mapping the root cause of the abnormality to the control measures. For heat transfer abnormalities where sublimation is hindered but the structure remains stable, the comparison between the rate offset and the preset offset threshold is used. Since the rate is low at this time, when the absolute value of the rate offset is greater than the preset offset threshold, it indicates that the sublimation driving force is seriously insufficient, and the heating power needs to be increased to enhance the heat input, thereby increasing the ice crystal sublimation rate. The adjustment range is calculated based on the product of the relative deviation and the preset first adjustment coefficient, achieving proportional control where the larger the deviation, the stronger the compensation. For structural collapse caused by hindered sublimation and structural deterioration, since the material has already suffered structural damage, simply increasing the heat will exacerbate the collapse. Therefore, it is necessary to reduce the heating power to weaken the thermal shock and at the same time reduce the conveying speed to extend the residence time in the pre-freezing zone, thereby strengthening the material skeleton from the root. The two control modes of energy supply and structural protection are decoupled, and the control strategy is automatically switched by identifying the abnormal type. This avoids under-adjustment when there is abnormal heat transfer and over-adjustment when there is structural collapse, thus achieving adaptive optimization control of the freeze-drying process.

[0088] Specifically, the second control unit includes:

[0089] A collapse deviation calculation subunit is used to calculate the collapse deviation based on the surface collapse degree and a preset collapse threshold.

[0090] A speed adjustment subunit, connected to the collapse deviation calculation subunit, is used to reduce the transmission speed when the collapse deviation is greater than a preset collapse deviation threshold, based on the relative deviation between the collapse deviation and the preset collapse deviation threshold and a preset speed adjustment coefficient, where T'=T×(1-b×|M-M'| / M'), T' is the adjusted transmission speed, T is the transmission speed, b is the preset speed adjustment coefficient, M is the collapse deviation, and M' is the preset collapse deviation threshold.

[0091] A power adjustment subunit, connected to the collapse deviation calculation subunit, is used to reduce the heating power based on the relative deviation between the collapse deviation and the preset collapse deviation threshold and a preset power adjustment coefficient when the collapse deviation is greater than a preset collapse deviation threshold. Here, E'=E×(1-c×|P-P'| / P'), E' is the adjusted heating power, E is the heating power, c is the preset power adjustment coefficient, P is the collapse deviation, and P' is the preset collapse deviation threshold.

[0092] The preset collapse threshold is a benchmark value used to determine whether unacceptable collapse has occurred on the surface of the material. It depends on the structural characteristics of high-pectin fruits, the product appearance quality grade requirements, and the calibration benchmark of the visual measurement system. It is usually set between 0.2 and 0.4. In this embodiment, it is set to 0.25. Considering that high-pectin fruits are prone to collapse but the initial small collapse does not affect the final quality, this threshold can trigger timely control when obvious structural damage occurs, while allowing a certain degree of small surface undulations, thus avoiding frequent intervention due to excessive sensitivity.

[0093] The preset collapse deviation threshold is used to determine whether the degree of collapse has reached the benchmark value that requires the initiation of regulation. It depends on the system's tolerance for the degree of collapse, the response characteristics of the regulation mechanism, and the need to avoid frequent adjustments. It is usually set between 10% and 30%. In this embodiment, it is set to 0.05, which filters out cases that have just exceeded the collapse threshold but are only slightly severe, avoids frequent actions of the actuator, and ensures that regulation is only initiated when the collapse reaches a certain severity, thus achieving optimal allocation of control resources.

[0094] The preset speed adjustment coefficient is a proportional coefficient used to control the adjustment range of the conveying speed. It depends on the response capability of the conveying system, the required residence time of the material in the pre-freezing zone, and the requirement to avoid excessive speed reduction affecting the output. It is usually set between 0.1 and 0.5. In this embodiment, it is set to 0.3, which can both extend the pre-freezing time by moderately reducing the speed to enhance structural stability and avoid a sharp drop in output or mismatch of subsequent drying time due to excessive speed reduction, thus achieving a balance between structural protection and production efficiency.

[0095] The preset power adjustment coefficient is a proportional coefficient used to control the adjustment range of heating power. It depends on the response capability of the heating system, the thermal sensitivity of the material, and the need to avoid the impact of a sudden drop in temperature on the drying process. It is usually set between 0.2 and 0.6. In this embodiment, it is set to 0.4, which can both slow down the collapse by moderately reducing the heating power and avoid the sublimation rate from dropping too sharply and the drying cycle from being excessively prolonged due to the power drop, thus achieving a balance between structural protection and drying efficiency.

[0096] A graded response mechanism for collapse deviation is used to achieve coordinated control of structural collapse anomalies. The degree of surface collapse deviation is used as the trigger for control intensity. Collapse deterioration is prevented through a dual approach of deceleration to strengthen the structure and cooling to reduce heat load. The real-time surface collapse degree is compared with a preset collapse threshold to calculate the collapse deviation, which directly quantifies the degree of structural damage exceeding the limit. When the collapse deviation exceeds the preset threshold, it indicates that the collapse has exceeded the allowable range and immediate intervention is required. This involves reducing the conveying speed to extend the material's residence time in the pre-freezing zone, promoting ice crystal refinement and skeleton strengthening, thereby fundamentally improving the structure's heat resistance. It also involves reducing heating power to decrease heat input to the sublimation zone, preventing high temperatures from exacerbating collapse. Furthermore, a proportional control mechanism with larger deviations and stronger deceleration and cooling amplitudes transforms the degree of structural damage into a linear mapping of deceleration and cooling intensity. Through dual-parameter coordinated adjustment, the further development of the current collapse is curbed, and the material's anti-collapse ability is fundamentally improved, achieving a comprehensive solution for structural collapse anomalies.

[0097] Specifically, the verification module includes:

[0098] A change calculation unit is used to calculate the change rate of the entire rehydration pass rate within the preset verification time to obtain the pass rate change rate;

[0099] The fluctuation calculation unit is used to calculate the standard deviation of the rehydration pass rate within the preset verification period to obtain the pass rate fluctuation value.

[0100] A verification unit, which is connected to the change calculation unit and the fluctuation calculation unit respectively, is used to issue an early warning when the pass rate change rate is less than a preset pass change threshold and the pass rate fluctuation value is greater than a preset pass fluctuation threshold.

[0101] The preset verification duration is the time length used to evaluate the control effect. It depends on the remaining transmission time of the material from the control point to the discharge area, the sampling frequency of the rehydration qualification rate detection, and the minimum sample size requirement to ensure statistical significance. It is usually set between 1 hour and 4 hours. In this embodiment, it is set to 2 hours, which can ensure that a sufficient number of rehydration qualification rate samples are collected for statistical calculation, and can issue an early warning in time when the control effect deviates, so as to avoid the continuous production of problematic batches.

[0102] The preset qualified change threshold is a threshold value used to determine whether the rehydration qualified rate tends to be stable. It depends on the normal fluctuation range of the rehydration qualified rate, the product quality stability requirements, and the statistical significance level. It is usually set between 0.01 per hour and 0.05 per hour. In this embodiment, it is set to 0.02 per hour, which not only recognizes the random fluctuations in normal production, but also identifies the continuous upward or downward trend caused by improper control, so as to achieve an accurate judgment on the effectiveness of control.

[0103] The preset acceptable fluctuation threshold is a threshold value used to judge the consistency of the rehydration qualification rate. It depends on the product's consistency requirements, the fluctuation level under normal production conditions, and the process control capability. It is usually set between 0.03 and 0.08. In this embodiment, it is set to 0.05, which allows a certain degree of normal fluctuation while identifying the increased instability caused by regulation, ensuring that the consistency within the product batch meets the standard.

[0104] An objective evaluation of the control effect is achieved through a dual verification mechanism. The freeze-drying process verification is deconstructed into two orthogonal dimensions: trend effectiveness and stability reliability. The combined characteristics of these two dimensions are used to determine whether the control truly solves the problem. The linear change slope of the rehydration pass rate within the verification period is calculated to quantify the improvement trend of quality after control. If the change rate of the pass rate is less than the preset pass rate change threshold, it indicates that the control has failed to effectively improve the rehydration performance of the product. The standard deviation of the rehydration pass rate is calculated to characterize the consistency level of quality. If the fluctuation value of the pass rate is greater than the preset fluctuation threshold, it indicates that the product quality is unstable and the control effect is unreliable. An early warning is issued when both exceed the standard. When the change rate of the pass rate is less than the threshold and the fluctuation value is greater than the threshold, it means that the control has neither improved the quality nor reduced the product consistency. This is a typical control failure mode, and manual intervention must be notified immediately. By combining the two independent evaluation criteria of whether the quality has improved and whether the quality is stable, the dual constraints effectively avoid the misjudgment that may be caused by a single indicator, and achieve a comprehensive and objective evaluation of the control effect. This provides a reliable decision-making basis for subsequent process library correction and threshold adaptive adjustment.

[0105] The adjustment module includes:

[0106] An anomaly density calculation unit is used to calculate the ratio of the timestamps of all freeze-drying anomalies within the preset observation period to the preset observation period, thereby obtaining the anomaly density.

[0107] An adjustment unit, connected to the abnormal density calculation unit, is used to increase the preset quality threshold based on the relative deviation between the abnormal density and the preset abnormal density threshold and a preset second adjustment coefficient when the abnormal density is greater than the preset abnormal density threshold. Here, H'=H×(1+d×|S-S'| / S'), H' is the adjusted preset quality threshold, H is the preset quality threshold, d is the preset second adjustment coefficient, S is the abnormal density, and S' is the preset abnormal density threshold.

[0108] The preset observation duration is the length of time used to statistically analyze the anomaly density. It depends on the operational stability of the production line, the frequency of abnormal events, and the system's response speed requirements for threshold adjustments. It is usually set between 24 hours and 7 days. In this embodiment, it is set to 72 hours, which can include enough production batches to reflect the statistical regularity of abnormal events and capture the changing trend of anomaly density in a timely manner, avoiding random fluctuation interference caused by an excessively short window or response lag caused by an excessively long window.

[0109] The preset anomaly density threshold is a threshold value used to determine whether the current anomaly occurrence frequency is too high. It depends on the acceptable frequency of abnormal events in the production process, quality control requirements, and the balance between false alarms and false negatives in the system. It is usually set between 0.1 times / hour and 0.3 times / hour. In this embodiment, it is set to 0.15 times / hour, which can identify abnormal states with significantly increased anomaly density, while avoiding frequent threshold adjustments due to short-term fluctuations, thus achieving a balance between detection sensitivity and production stability.

[0110] The preset adjustment coefficient is a proportional coefficient used to control the adjustment range of the quality threshold. It depends on the sensitivity of the quality threshold to anomaly detection, the system's tolerance to false alarms and false negatives, and the need to avoid threshold oscillations. It is usually set between 0.1 and 0.5. In this embodiment, it is set to 0.3. This can effectively reduce frequent anomaly alarms caused by oversensitivity by moderately increasing the quality threshold, and avoid the risk of false negatives caused by the threshold deviating from the normal range due to excessive adjustment range. This achieves a smooth transition of anomaly detection adaptive optimization.

[0111] The dynamic optimization of detection sensitivity is achieved through an anomaly density-driven adaptive quality threshold adjustment mechanism. Its underlying logic involves using the frequency characteristics of historical anomalies as feedback signals for threshold optimization, and correcting the judgment criteria by quantifying the deviation between the anomaly density and the preset threshold. The timestamp density of anomalies within a preset observation period characterizes the current anomaly triggering frequency of the system. When the anomaly density exceeds the preset anomaly density threshold, it indicates that the current quality threshold is too strict, leading to frequent anomaly triggers. In this case, the quality threshold is increased, thereby reducing system sensitivity and unnecessary intervention. The density of historical anomalies serves as a negative feedback signal for threshold optimization. By dynamically balancing detection sensitivity and system stability, the quality threshold is always kept at its optimal operating point. This avoids frequent false alarms that disrupt production due to an excessively low threshold, while also preventing missed anomalies due to an excessively high threshold, ultimately achieving adaptive optimization and long-term stable operation of anomaly detection performance.

[0112] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A freeze-drying parameter intelligent control system, characterized in that, include: The data acquisition module is used to collect data in real time on each material to be tested on the freeze-drying production line, including the ice crystal migration rate and surface collapse degree in the sublimation drying zone, the color difference change rate in the desorption drying zone, the rehydration rate in the discharge zone, and the rehydration qualification rate. The determination module is used to determine the occurrence of a freeze-drying abnormality event based on the comparison result between the quality index and the preset quality threshold. The quality index is calculated by weighting the surface collapse degree, the color difference change rate and the rehydration rate. A stage determination module is used to determine the occurrence of sublimation drying anomaly based on the freeze-drying anomaly event and according to the numerical characteristics of the changing coupling degree, wherein the changing coupling degree is determined based on the temporal variation characteristics of the ice crystal migration rate and the surface collapse degree. The type determination module is used to determine the anomaly type as heat transfer anomaly or structural collapse anomaly based on the sublimation drying anomaly and according to the temporal correlation characteristics of the ice crystal migration rate and the surface collapse degree. The control module is used to adjust the heating power of the sublimation drying zone according to the ice crystal moving rate based on the heat transfer anomaly type, and to adjust the conveying speed and heating power according to the surface collapse degree based on the structural collapse type. The verification module is used to perform freeze-drying verification based on the adjusted heating power and the conveying speed, according to the change characteristics of the rehydration qualification rate within a preset verification time, and to issue an early warning based on the verification results. The adjustment module is used to adjust the preset quality threshold according to the temporal characteristics of the freeze-drying abnormal event occurring within the preset observation period. The variable coupling degree is calculated based on the directional synchronization factor, nonlinear deviation degree, preset directional weight, and preset nonlinear weight. The directional synchronization factor is calculated based on the cosine similarity between the ice crystal pushing rate and the surface collapse degree, and the nonlinear deviation degree is calculated based on the Pearson correlation coefficient between the ice crystal pushing rate and the surface collapse degree. The type determination module includes: The feature calculation unit is used to calculate the rate offset based on all the ice crystal pushing rates and the preset target rate within a preset time period, and to calculate the collapse change rate based on all the surface collapse degrees within a preset time period. A type determination unit, connected to the feature calculation unit, is used to determine the anomaly type as the heat transfer anomaly type when the rate offset is less than a first preset threshold and the collapse change rate is less than or equal to a second preset threshold, and to determine the anomaly type as the structural collapse type when the rate offset is less than the first preset threshold and the collapse change rate is greater than the second preset threshold.

2. The intelligent control system for freeze-drying parameters according to claim 1, characterized in that, When the quality index is less than the preset quality threshold, a freeze-drying abnormality event is determined to have occurred.

3. The intelligent control system for freeze-drying parameters according to claim 2, characterized in that, The quality index is calculated by weighting the surface collapse degree, the color difference change rate, the rehydration rate, the preset collapse weight, the preset color difference weight, and the preset rehydration weight.

4. The intelligent control system for freeze-drying parameters according to claim 3, characterized in that, When the changed coupling degree is greater than a preset coupling degree threshold, a sublimation drying anomaly is determined to have occurred.

5. The intelligent control system for freeze-drying parameters according to claim 4, characterized in that, The control module includes: The first control unit is used to calculate the rate offset based on the heat transfer anomaly type, the ice crystal pushing rate and the preset target rate, and adjust the heating power based on the comparison result of the rate offset and the preset offset threshold. The second control unit is used to adjust the conveying speed and the heating power based on the surface collapse degree, according to the structure collapse type.

6. The intelligent control system for freeze-drying parameters according to claim 5, characterized in that, The second control unit includes: A collapse deviation calculation subunit is used to calculate the collapse deviation based on the surface collapse degree and a preset collapse threshold. A speed adjustment subunit, which is connected to the collapse deviation calculation subunit, is used to adjust the transmission speed according to the comparison result of the collapse deviation and the preset collapse deviation threshold and the preset speed adjustment coefficient. A power adjustment subunit, which is connected to the collapse deviation calculation subunit, is used to adjust the heating power according to the comparison result of the collapse deviation and the preset collapse deviation threshold and the preset power adjustment coefficient.

7. The intelligent control system for freeze-drying parameters according to claim 6, characterized in that, The verification module includes: A change calculation unit is used to calculate the change rate of the pass rate based on all the rehydration pass rates within the preset verification time. A fluctuation calculation unit is used to calculate the fluctuation value of the pass rate based on the pass rate of all the rehydration pass rates within the preset verification period. A verification unit, which is connected to the change calculation unit and the fluctuation calculation unit respectively, is used to issue an early warning when the pass rate change rate is less than a preset pass change threshold and the pass rate fluctuation value is greater than a preset pass fluctuation threshold.

8. The intelligent control system for freeze-drying parameters according to claim 7, characterized in that, The adjustment module includes: An anomaly density calculation unit is used to calculate the anomaly density based on the timestamps of all freeze-drying anomalies within the preset observation period. An adjustment unit, connected to the abnormal density calculation unit, is used to adjust the preset quality threshold based on the comparison result between the abnormal density and the preset abnormal density threshold.

Images