Real-time monitoring optimization system for pharmaceutical sterilization process and method thereof

Abstract

The application discloses a real-time monitoring optimization system and method in a drug sterilization process, and belongs to the technical field of drug production. The method acquires real-time physical state information of a container by collecting thermal imaging scanning data and vibration fluctuation signals, extracts edge contour geometric features to determine the degree of deformation by using an image processing algorithm, and if the degree of deformation exceeds a threshold value, analyzes abnormal fluctuation features of the deformation area by using a deep learning model, matches the vibration signal to determine an internal pressure imbalance area, and then generates a comprehensive influence vector by fusion to extract key dimensions to obtain a sterilization effect prediction score. When the sterilization effect prediction score is lower than a threshold value, the process parameters are adjusted according to the vector, and the reliability is verified through a simulation module, so that precise control and quality assurance of the sterilization process are finally realized. The method significantly improves sterilization reliability and production batch consistency, quantifies the drug quality level, and ensures the safety and efficiency of the pharmaceutical process.

Description

Technical Field

[0001] This invention relates to the field of pharmaceutical manufacturing technology, specifically to a real-time monitoring and optimization system and method for the sterilization process of pharmaceuticals. Background Technology

[0002] Sterilization of intravenous medications is a crucial step in pharmaceutical manufacturing, directly impacting drug safety and patient health. Research and practice in this area are irreplaceable in ensuring drug quality and preventing infection risks. Especially in the production of intravenous drugs, the stability and reliability of sterilization effectiveness are considered the cornerstone of the industry's development. Even minor errors can lead to serious consequences; therefore, exploring more efficient and controllable sterilization methods is an urgent need.

[0003] However, current mainstream sterilization methods have revealed some deep-seated shortcomings in practical applications. Many technologies rely too heavily on preset process parameters and indirect detection methods, lacking a direct understanding of the pharmaceutical container and its internal state during sterilization. This approach often fails to capture subtle changes in a timely manner, such as minor deformations of the container under high temperature and pressure or abnormal fluctuations in the liquid state, leading to the overlooking of potential problems and consequently affecting the safety and batch consistency of the pharmaceutical products.

[0004] A deeper technological challenge lies in the fact that sterilization is a dynamic and complex environment involving various physical changes under high temperature and pressure, and existing technologies have significant shortcomings in their ability to observe and assess these changes in real time. In particular, the subtle deformations that may occur in pharmaceutical containers during sterilization can further destabilize the internal liquid state, such as the abnormal generation of bubbles or uneven liquid distribution. These factors interact, making it difficult to accurately predict and control the sterilization effect. For example, in some high-temperature sterilization scenarios, slight deformation of the container edges due to uneven heating can lead to an imbalance in internal pressure distribution, resulting in localized boiling or bubble accumulation. If this situation is not detected in time, it may directly affect the final quality of the pharmaceutical product.

[0005] Therefore, how to capture subtle changes in the state of drug containers in real time under complex high temperature and high pressure environments, and accurately determine the impact of these changes on sterilization effect, has become a key issue in improving the reliability and safety of intravenous drug sterilization. Summary of the Invention

[0006] This invention proposes a real-time monitoring and optimization system and method for the sterilization process of pharmaceuticals to solve the above-mentioned technical problems.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0008] A method for real-time monitoring and optimization during drug sterilization, the method comprising:

[0009] Multimodal monitoring data is collected from the surface of a pharmaceutical container in a pharmaceutical sterilization environment. The multimodal monitoring data includes thermal imaging scan data and vibration fluctuation signals.

[0010] The system acquires real-time physical state information of a drug container under high temperature and high pressure conditions, analyzes the thermal imaging scan data, extracts the geometric features of the drug container's edge contour, calculates the degree of subtle deformation on the surface of the drug container, and determines potential deformation locations.

[0011] If the degree of minor deformation on the surface of the drug container exceeds the preset safety threshold, a detailed analysis is performed on the potential deformation location area. The image data of the potential deformation location area is processed by a trained deep learning model to obtain the abnormal fluctuation characteristics of the internal distribution of the drug liquid and determine the reaction state of the liquid during the sterilization process.

[0012] Based on the abnormal fluctuation characteristics, the vibration fluctuation signal is decomposed and matched with the abnormal fluctuation characteristics to determine the stability of the liquid state of the drug.

[0013] Based on the stability of the liquid state of the drug and the degree of minor deformation on the surface of the drug container, a comprehensive influence vector is generated. Key dimensions are extracted from the comprehensive influence vector, and a sterilization effect prediction score is calculated to quantify the drug quality assurance level.

[0014] If the sterilization effect prediction score is lower than the preset quality threshold, the sterilization process parameters are adjusted according to the comprehensive influence vector to generate an optimized process setting.

[0015] By inputting the current physical state information of the drug container into a multiphysics simulation model, the optimized process settings are verified, the degree of reliability improvement in the sterilization process is judged, and the final drug sterilization control scheme is determined.

[0016] As a preferred embodiment of the real-time monitoring and optimization method for the sterilization process of a drug in this invention, the specific implementation process for calculating the degree of minute deformation on the surface of the drug container includes:

[0017] The system acquires real-time physical state information of the drug container under high temperature and high pressure conditions, and analyzes the thermal imaging scan data using an image processing algorithm to identify multiple feature points on the edge contour of the drug container.

[0018] For each feature point, calculate the deviation between its current coordinate position and the initial reference coordinate position. The calculation formula is as follows:

[0019] ;

[0020] in, This represents the deviation value of the i-th feature point. , This represents the current coordinate position of the i-th feature point in the thermal imaging scan. , This represents the initial reference coordinate position of the i-th feature point;

[0021] Based on the deviation value and the weighting coefficient of each feature point, the degree of subtle deformation on the surface of the drug container is comprehensively evaluated. The evaluation formula is as follows:

[0022] ;

[0023] in, This indicates the degree of minute deformation on the surface of the drug container, where M represents the total number of feature points on the container's edge contour. This represents the weight coefficient of the i-th feature point, used to reflect the importance of different regional blocks to the overall deformation assessment;

[0024] The container surface is divided into several regions, and for each region r, the average offset of all feature points within the region is calculated:

[0025]

[0026] The standard deviation of the offset within the region, used to determine whether the deformation is uniform, is calculated based on the average offset of all feature points within the region. ;

[0027] When the average offset of all feature points in a region is less than the offset threshold and the standard deviation is less than the standard threshold, the region is marked as a potential deformation location region.

[0028] If multiple adjacent regions are simultaneously potential deformation locations, they are merged into one potential deformation location region.

[0029] As a preferred embodiment of the real-time monitoring and optimization method for the sterilization process of a drug in this invention, the acquisition of abnormal fluctuation characteristics of the internal distribution of the drug liquid includes:

[0030] The image data of the potential deformation location region is input into the trained deep learning model;

[0031] The image data is used to extract features through a deep learning model to identify abnormal distribution points inside the liquid medicine.

[0032] Based on the distribution anomalies, the fluctuation characteristics of the drug liquid during the sterilization process are analyzed to generate corresponding reaction state description data;

[0033] Based on the reaction state description data, the reaction states of the drug liquid are classified, and their degree of abnormality is marked as the abnormal fluctuation characteristics of the drug liquid.

[0034] As a preferred embodiment of the real-time monitoring and optimization method for drug sterilization process in this invention, the method involves decomposing the vibration fluctuation signal based on the abnormal fluctuation characteristics and matching it with the abnormal fluctuation characteristics to determine the stability of the drug liquid state, including:

[0035] The vibration wave signal is decomposed into frequency components using signal analysis methods, and the amplitude characteristics of each component are extracted.

[0036] The amplitude characteristics are compared with the abnormal fluctuation characteristics of the drug liquid to assess the overall stability of the drug liquid state. The assessment formula is as follows:

[0037]

[0038] in, The overall stability score of the drug in its liquid state is represented by K, where K represents the number of frequency components after decomposition of the vibration signal. This represents the amplitude value of the k-th frequency component. Indicates the standard reference amplitude. Indicates the maximum permissible amplitude. The standard deviation of the vibration wave signal is represented. The threshold representing the reference standard deviation of the vibration wave signal.

[0039] As a preferred embodiment of the real-time monitoring and optimization method for drug sterilization in this invention, a comprehensive influence vector is generated based on the stability of the drug liquid state and the degree of minor deformation on the surface of the drug container. Key dimensions are extracted from the comprehensive influence vector, and a sterilization effect prediction score is calculated to quantify the drug quality assurance level, including:

[0040] The stability of the liquid state of the drug and the degree of minor deformation on the surface of the drug container are weighted and combined to generate a comprehensive influence vector.

[0041] Key dimension data are extracted from the comprehensive influence vector, and the sterilization effect prediction score is calculated using the following formula:

[0042]

[0043] in, Indicates the sterilization effectiveness prediction score. Weighting coefficients representing liquid stability Weighting coefficients representing the degree of container deformation. Weighting coefficients representing abnormal fluctuation characteristics. The safety threshold indicating the degree of deformation This represents the intensity value of abnormal fluctuation features in the internal distribution of drug liquid extracted by a deep learning model.

[0044] As a preferred embodiment of the real-time monitoring and optimization method for drug sterilization in this invention, if the sterilization effect prediction score is lower than a preset quality threshold, the sterilization process parameters are adjusted according to the comprehensive influence vector to generate an optimized process setting, including:

[0045] Based on the key dimension data of the comprehensive influence vector, the sterilization process parameters are adjusted according to the preset parameter adjustment rules;

[0046] Based on the adjusted parameters, an optimized process setting is generated.

[0047] As a preferred embodiment of the real-time monitoring and optimization method for the sterilization process of pharmaceuticals in this invention, the current physical state information of the pharmaceutical container is input into a multi-physics simulation model to verify the optimized process settings, determine the degree of reliability improvement in the sterilization process, and determine the final pharmaceutical sterilization control scheme, including:

[0048] A multiphysics simulation model of a drug container-liquid-sterilization environment is constructed based on the finite element method, with the current physical state data of the container as input.

[0049] Load the optimized process settings, simulate the entire sterilization process, and output key node data, including liquid stability, maximum minor deformation of the container, and fluctuation amplitude of the drug liquid. Calculate the reliability improvement of the optimized scheme. The calculation formula is:

[0050]

[0051] in, This represents the predicted score for the optimized sterilization effect. This represents the predicted sterilization effect score before optimization.

[0052] When the improvement in reliability exceeds the reliability threshold, the final drug sterilization control plan is generated.

[0053] A real-time monitoring and optimization system for the sterilization process of pharmaceuticals, the system comprising a data acquisition module, a container analysis module, a liquid internal analysis module, a liquid state determination module, a sterilization effect prediction module, an optimization module, and a control scheme generation module;

[0054] The data acquisition module is used to collect multimodal monitoring data from the surface of a drug container in a drug sterilization environment. The multimodal monitoring data includes thermal imaging scan data and vibration wave signals.

[0055] The container analysis module is used to acquire real-time physical state information of the drug container under high temperature and high pressure conditions, analyze the thermal imaging scan data, extract the geometric features of the drug container edge contour, calculate the degree of subtle deformation on the surface of the drug container, and determine the potential deformation location area.

[0056] The liquid internal analysis module is used to perform detailed analysis on potential deformation locations. It processes the image data of potential deformation locations using a trained deep learning model to obtain abnormal fluctuation characteristics of the internal distribution of the drug liquid and determine the reaction state of the liquid during the sterilization process.

[0057] The liquid state determination module is used to decompose the vibration fluctuation signal and match it with the abnormal fluctuation characteristics based on the abnormal fluctuation characteristics to determine the stability of the liquid state of the drug.

[0058] The sterilization effect prediction module is used to generate a comprehensive influence vector based on the stability of the liquid state of the drug and the degree of minor deformation on the surface of the drug container. Key dimensions are extracted from the comprehensive influence vector to calculate the sterilization effect prediction score and quantify the drug quality assurance level.

[0059] The optimization module is used to adjust the sterilization process parameters according to the comprehensive influence vector and generate optimized process settings;

[0060] The control scheme generation module is used to input the current physical state information of the drug container through a multiphysics simulation model, verify the optimized process settings, judge the degree of reliability improvement of the sterilization process, and determine the final drug sterilization control scheme.

[0061] Compared with existing technologies, the beneficial effects achieved by this invention are as follows: The real-time monitoring and optimization system and method for pharmaceutical sterilization provided by this invention addresses the core business problem of containers in the pharmaceutical sterilization environment being prone to slight deformation under high temperature and high pressure conditions, leading to unstable internal liquid states and affecting sterilization effectiveness and pharmaceutical quality consistency. By collecting thermal imaging scanning data and vibration fluctuation signals, real-time physical state information of the container is obtained. Image processing algorithms are used to extract geometric features of the edge contours to determine the degree of deformation. If the deformation exceeds a threshold, a deep learning model is used to analyze the abnormal fluctuation characteristics of the deformed area, matching them with the vibration signal to identify areas of internal pressure imbalance. This is then fused to generate a comprehensive influence vector, extracting key dimensions to obtain a sterilization effectiveness prediction score. When the score is below a threshold, process parameters are adjusted based on the vector, and reliability is verified through a simulation module. Ultimately, precise control and quality assurance of the sterilization process are achieved. This method significantly improves sterilization reliability and production batch consistency, quantifies pharmaceutical quality levels, and ensures the safety and efficiency of the pharmaceutical manufacturing process. Attached Figure Description

[0062] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0063] Figure 1 This is a schematic diagram of the method steps in an embodiment of the present invention;

[0064] Figure 2 This is a schematic diagram of the system structure in an embodiment of the present invention. Detailed Implementation

[0065] 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.

[0066] Please see Figures 1-2 In this first embodiment: a real-time monitoring and optimization method for the sterilization process of pharmaceuticals, the method includes:

[0067] Multimodal monitoring data was collected from the surface of drug containers in a drug sterilization environment. The multimodal monitoring data included thermal imaging scan data and vibration fluctuation signals.

[0068] An infrared thermal imager is deployed at the viewing window on the inner wall of the sterilizer, covering the array of medicine containers with a sampling frequency of no less than 10 Hz and a temperature resolution of less than or equal to 0.05℃, to acquire images of the surface temperature distribution and thermal radiation of the containers. A piezoelectric accelerometer is attached to the support of the medicine containers or directly contacts the bottom of the containers through a high-temperature resistant coupling agent, with a sampling frequency of 1kHz to 10kHz, to acquire micro-vibration signals caused by internal liquid fluctuations and gas releases under high temperature and high pressure. A time synchronization module, such as GPS or IEEE 1588 protocol, ensures the alignment of thermal imaging data and vibration signals on the time axis, facilitating subsequent fusion analysis. The thermal imaging images are denoised and contrast enhanced. The vibration signals are filtered, such as by bandpass filtering from 0.5 Hz to 2 kHz, to remove the mechanical noise of the sterilizer itself.

[0069] The system acquires real-time physical state information of drug containers under high temperature and high pressure conditions, analyzes thermal imaging scan data, extracts geometric features of the drug container edge contour, calculates the degree of subtle deformation on the surface of the drug container, and determines potential deformation locations.

[0070] If the degree of minor deformation on the surface of the medicine container exceeds the preset safety threshold, a detailed analysis is performed on the potential deformation location area. The image data of the potential deformation location area is processed by a trained deep learning model to obtain the abnormal fluctuation characteristics of the internal distribution of the medicine liquid and determine the reaction state of the liquid during the sterilization process.

[0071] Based on the abnormal fluctuation characteristics, the vibration fluctuation signal is decomposed and matched with the abnormal fluctuation characteristics to determine the stability of the drug liquid state.

[0072] Based on the stability of the liquid state of the drug and the degree of minor deformation on the surface of the drug container, a comprehensive influence vector is generated. Key dimensions are extracted from the comprehensive influence vector, and the sterilization effect prediction score is calculated to quantify the drug quality assurance level.

[0073] If the sterilization effect prediction score is lower than the preset quality threshold, the sterilization process parameters are adjusted according to the comprehensive influence vector to generate an optimized process setting.

[0074] By inputting the current physical state information of the drug container into a multiphysics simulation model, the optimized process settings are verified, the degree of reliability improvement in the sterilization process is judged, and the final drug sterilization control scheme is determined.

[0075] Specifically, the implementation process for calculating the degree of minute deformation on the surface of a drug container includes:

[0076] The system acquires real-time physical state information of drug containers under high temperature and high pressure conditions, analyzes thermal imaging scan data using image processing algorithms, and identifies multiple feature points on the edge contour of the drug container. The Canny edge detection algorithm can be used to extract the container edge contour, and the Shi-Tomasi corner detection algorithm can be used to extract key feature points on the contour.

[0077] A reference coordinate system is established using images from the initial state (room temperature and pressure before sterilization). Subpixel-level matching algorithms (such as optical flow or SIFT) are used to calculate the coordinate offset of feature points in the current frame. Specifically, for each feature point, the deviation between its current coordinate position and the initial reference coordinate position is calculated using the following formula:

[0078] ;

[0079] in, This represents the deviation value of the i-th feature point. , This represents the current coordinate position of the i-th feature point in the thermal imaging scan. , This represents the initial reference coordinate position of the i-th feature point;

[0080] Based on the deviation value and the weighting coefficient of each feature point, the degree of subtle deformation on the surface of the drug container is comprehensively evaluated. The evaluation formula is as follows:

[0081] ;

[0082] in, This indicates the degree of minute deformation on the surface of the drug container, where M represents the total number of feature points on the container's edge contour. This represents the weight coefficient of the i-th feature point used to reflect the importance of different regions to the overall deformation assessment. It is set according to the importance of the container structure, such as the bottle mouth, bottle bottom, and middle of the bottle body having higher weights.

[0083] The container surface is divided into several regions, and for each region r, the average offset of all feature points within the region is calculated:

[0084]

[0085] The standard deviation of the offset within the region, used to determine whether the deformation is uniform, is calculated based on the average offset of all feature points within the region. ;

[0086] When the average offset of all feature points in a region is less than the offset threshold and the standard deviation is less than the standard threshold, the region is marked as a potential deformation location region.

[0087] If multiple adjacent regions are simultaneously potential deformation locations, they are merged into one potential deformation location region.

[0088] Specifically, obtaining abnormal fluctuation characteristics of the internal distribution of drug liquids includes:

[0089] Image data of potential deformation locations are input into a trained deep learning model. Lightweight convolutional neural networks, such as MobileNetV3 or EfficientNet-Lite, can be used and deployed on edge computing devices, such as embedded GPUs, to ensure real-time performance. Deformation area images collected during historical sterilization processes are manually labeled to indicate whether there are bubbles, stratification, or boiling anomalies inside the liquid, and a deep learning model is constructed. The model outputs the degree of anomaly and combines it with heatmaps to locate abnormal areas.

[0090] By using a deep learning model to extract features from image data, abnormal distribution points inside the liquid medicine can be identified.

[0091] Based on the distribution of outliers, analyze the fluctuation characteristics of the drug liquid during the sterilization process and generate corresponding reaction state description data;

[0092] Based on the reaction state description data, the reaction states of drug liquids are classified, and their degree of abnormality is marked as a characteristic of abnormal fluctuations in drug liquids.

[0093] Specifically, based on the abnormal fluctuation characteristics, the vibration fluctuation signal is decomposed and matched with the abnormal fluctuation characteristics to determine the stability of the drug liquid state, including:

[0094] The vibration wave signal is decomposed into frequency components by signal analysis methods to extract the amplitude characteristics of each component; wavelet packet decomposition or empirical mode decomposition can be used to decompose the vibration signal into multiple intrinsic mode functions, extract the energy distribution of each frequency band, and obtain the amplitude characteristics of each component.

[0095] Correlation analysis (e.g., cross-correlation function) is performed between the frequency band corresponding to the abnormal fluctuation characteristics (e.g., 100 Hz ~ 500 Hz) and the vibration components to locate areas of uneven pressure distribution. This involves comparing the amplitude characteristics with the abnormal fluctuation characteristics of the drug liquid to assess the overall stability of the drug liquid state. The assessment formula is as follows:

[0096]

[0097] in, The overall stability score of the drug in its liquid state ranges from 0 to 1, with values ​​closer to 1 indicating greater stability. K represents the number of frequency components after decomposition of the vibration signal. This represents the amplitude value of the k-th frequency component. Indicates the standard reference amplitude. Indicates the maximum permissible amplitude. The standard deviation of the vibration wave signal is represented. The threshold representing the reference standard deviation of the vibration wave signal.

[0098] Specifically, based on the stability of the liquid state of the drug and the degree of minor deformation on the surface of the drug container, a comprehensive influence vector is generated. Key dimensions are extracted from the comprehensive influence vector, and a sterilization effect prediction score is calculated to quantify the drug quality assurance level, including:

[0099] The stability of the liquid state of the drug and the degree of minor deformation on the surface of the drug container are weighted and combined to generate a comprehensive influence vector.

[0100] Key dimension data are extracted from the comprehensive influence vector, and the sterilization effect prediction score is calculated using the following formula:

[0101]

[0102] in, Indicates the sterilization effectiveness prediction score. Weighting coefficients representing liquid stability Weighting coefficients representing the degree of container deformation. Weighting coefficients representing abnormal fluctuation characteristics. The safety threshold indicating the degree of deformation This represents the intensity value of abnormal fluctuation features in the internal distribution of drug liquid extracted by a deep learning model.

[0103] Specifically, if the predicted sterilization effect score is lower than the preset quality threshold, the sterilization process parameters are adjusted based on the comprehensive influence vector to generate optimized process settings, including:

[0104] Based on the key dimension data of the comprehensive influence vector, adjust the sterilization process parameters according to the preset parameter adjustment rules; for example:

[0105] like If the temperature is too low, reduce the heating rate or extend the soaking time;

[0106] like If the temperature is too high, lower the maximum sterilization temperature or optimize the pressure matching.

[0107] Based on the adjusted parameters, an optimized process setting is generated.

[0108] Specifically, by inputting the current physical state information of the drug container into a multiphysics simulation model, the optimized process settings are verified, the degree of reliability improvement in the sterilization process is assessed, and the final drug sterilization control scheme is determined, including:

[0109] A multiphysics simulation model of a drug container-liquid-sterilization environment is constructed based on the finite element method, with the current physical state data of the container as input.

[0110] Load the optimized process settings, simulate the entire sterilization process, and output key node data, including liquid stability, maximum minor deformation of the container, and fluctuation amplitude of the drug liquid. Calculate the reliability improvement of the optimized scheme. The calculation formula is:

[0111]

[0112] in, This represents the predicted score for the optimized sterilization effect. This represents the predicted sterilization effect score before optimization.

[0113] When the improvement in reliability exceeds the reliability threshold, the final drug sterilization control plan is generated.

[0114] A real-time monitoring and optimization system for the sterilization process of pharmaceuticals, comprising a data acquisition module, a container analysis module, a liquid internal analysis module, a liquid state determination module, a sterilization effect prediction module, an optimization module, and a control scheme generation module;

[0115] The data acquisition module is used to collect multimodal monitoring data from the surface of drug containers in a drug sterilization environment. The multimodal monitoring data includes thermal imaging scan data and vibration fluctuation signals.

[0116] The container analysis module is used to acquire real-time physical state information of drug containers under high temperature and high pressure conditions, analyze thermal imaging scan data, extract geometric features of the drug container edge contour, calculate the degree of subtle deformation on the surface of the drug container, and determine potential deformation location areas.

[0117] The liquid internal analysis module is used to perform detailed analysis on potential deformation locations. It processes the image data of potential deformation locations using a trained deep learning model to obtain abnormal fluctuation characteristics of the internal distribution of the drug liquid and determine the reaction state of the liquid during the sterilization process.

[0118] The liquid state determination module is used to decompose the vibration fluctuation signal and match it with the abnormal fluctuation characteristics to determine the stability of the drug liquid state.

[0119] The sterilization effect prediction module is used to generate a comprehensive influence vector based on the stability of the liquid state of the drug and the degree of minor deformation on the surface of the drug container. Key dimensions are extracted from the comprehensive influence vector to calculate the sterilization effect prediction score and quantify the drug quality assurance level.

[0120] The optimization module is used to adjust sterilization process parameters based on the comprehensive influence vector and generate optimized process settings.

[0121] The control scheme generation module is used to input the current physical state information of the drug container through a multiphysics simulation model, verify the optimized process settings, judge the degree of reliability improvement of the sterilization process, and determine the final drug sterilization control scheme.

[0122] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0123] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for real-time monitoring and optimization during the sterilization process of pharmaceuticals, characterized in that, The method includes: Multimodal monitoring data is collected from the surface of a pharmaceutical container in a pharmaceutical sterilization environment. The multimodal monitoring data includes thermal imaging scan data and vibration fluctuation signals. The system acquires real-time physical state information of a drug container under high temperature and high pressure conditions, analyzes the thermal imaging scan data, extracts the geometric features of the drug container's edge contour, calculates the degree of subtle deformation on the surface of the drug container, and determines potential deformation locations. If the degree of minor deformation on the surface of the drug container exceeds the preset safety threshold, a detailed analysis is performed on the potential deformation location area. The image data of the potential deformation location area is processed by a trained deep learning model to obtain the abnormal fluctuation characteristics of the internal distribution of the drug liquid and determine the reaction state of the liquid during the sterilization process. Based on the abnormal fluctuation characteristics, the vibration fluctuation signal is decomposed and matched with the abnormal fluctuation characteristics to determine the stability of the liquid state of the drug. Based on the stability of the liquid state of the drug and the degree of minor deformation on the surface of the drug container, a comprehensive influence vector is generated. Key dimensions are extracted from the comprehensive influence vector, and a sterilization effect prediction score is calculated to quantify the drug quality assurance level. If the sterilization effect prediction score is lower than the preset quality threshold, the sterilization process parameters are adjusted according to the comprehensive influence vector to generate an optimized process setting. By inputting the current physical state information of the drug container into a multiphysics simulation model, the optimized process settings are verified, the degree of reliability improvement in the sterilization process is judged, and the final drug sterilization control scheme is determined.

2. The real-time monitoring and optimization method for the sterilization process of pharmaceuticals according to claim 1, characterized in that, The specific implementation process for calculating the degree of minute deformation on the surface of the drug container includes: The system acquires real-time physical state information of the drug container under high temperature and high pressure conditions, and analyzes the thermal imaging scan data using an image processing algorithm to identify multiple feature points on the edge contour of the drug container. For each feature point, calculate the deviation between its current coordinate position and the initial reference coordinate position. The calculation formula is as follows: ； in, This represents the deviation value of the i-th feature point. , This represents the current coordinate position of the i-th feature point in the thermal imaging scan. , This represents the initial reference coordinate position of the i-th feature point; Based on the deviation value and the weighting coefficient of each feature point, the degree of subtle deformation on the surface of the drug container is comprehensively evaluated. The evaluation formula is as follows: ； in, This indicates the degree of minute deformation on the surface of the drug container, where M represents the total number of feature points on the container's edge contour. This represents the weight coefficient of the i-th feature point, used to reflect the importance of different regional blocks to the overall deformation assessment; The container surface is divided into several regions, and for each region r, the average offset of all feature points within the region is calculated: ； The standard deviation of the offset within the region, used to determine whether the deformation is uniform, is calculated based on the average offset of all feature points within the region. ; When the average offset of all feature points in a region is less than the offset threshold and the standard deviation is less than the standard threshold, the region is marked as a potential deformation location region. If multiple adjacent regions are simultaneously potential deformation locations, they are merged into one potential deformation location region.

3. The real-time monitoring and optimization method for the sterilization process of pharmaceuticals according to claim 2, characterized in that, The abnormal fluctuation characteristics of the internal distribution of the drug liquid include: The image data of the potential deformation location region is input into the trained deep learning model; The image data is used to extract features through a deep learning model to identify abnormal distribution points inside the liquid medicine. Based on the distribution anomalies, the fluctuation characteristics of the drug liquid during the sterilization process are analyzed to generate corresponding reaction state description data; Based on the reaction state description data, the reaction states of the drug liquid are classified, and their degree of abnormality is marked as the abnormal fluctuation characteristics of the drug liquid.

4. The real-time monitoring and optimization method for the sterilization process of pharmaceuticals according to claim 3, characterized in that, The step of decomposing the vibration fluctuation signal based on the abnormal fluctuation characteristics and matching it with the abnormal fluctuation characteristics to determine the stability of the drug liquid state includes: The vibration wave signal is decomposed into frequency components using signal analysis methods, and the amplitude characteristics of each component are extracted. The amplitude characteristics are compared with the abnormal fluctuation characteristics of the drug liquid to assess the overall stability of the drug liquid state. The assessment formula is as follows: ； in, The overall stability score of the drug in its liquid state is represented by K, where K represents the number of frequency components after decomposition of the vibration signal. This represents the amplitude value of the k-th frequency component. Indicates the standard reference amplitude. Indicates the maximum permissible amplitude. The standard deviation of the vibration wave signal is represented. The threshold representing the reference standard deviation of the vibration wave signal.

5. The real-time monitoring and optimization method for the sterilization process of pharmaceuticals according to claim 4, characterized in that, The process involves generating a comprehensive influence vector based on the stability of the liquid drug and the degree of minor deformation on the surface of the drug container. Key dimensions are extracted from this comprehensive influence vector to calculate a sterilization effectiveness prediction score, quantifying the drug quality assurance level. This includes: The stability of the liquid state of the drug and the degree of minor deformation on the surface of the drug container are weighted and combined to generate a comprehensive influence vector. Key dimension data are extracted from the comprehensive influence vector, and the sterilization effect prediction score is calculated using the following formula: ； in, Indicates the sterilization effectiveness prediction score. Weighting coefficients representing liquid stability Weighting coefficients representing the degree of container deformation. Weighting coefficients representing abnormal fluctuation characteristics. The safety threshold indicating the degree of deformation This represents the intensity value of abnormal fluctuation features in the internal distribution of drug liquid extracted by a deep learning model.

6. The real-time monitoring and optimization method for the sterilization process of pharmaceuticals according to claim 5, characterized in that, If the predicted sterilization effect score is lower than a preset quality threshold, the sterilization process parameters are adjusted according to the comprehensive influence vector to generate an optimized process setting, including: Based on the key dimension data of the comprehensive influence vector, the sterilization process parameters are adjusted according to the preset parameter adjustment rules; Based on the adjusted parameters, an optimized process setting is generated.

7. The real-time monitoring and optimization method for the sterilization process of pharmaceuticals according to claim 6, characterized in that, The process of inputting the current physical state information of the drug container into a multiphysics simulation model to verify the optimized process settings, determine the degree of reliability improvement in the sterilization process, and determine the final drug sterilization control scheme includes: A multiphysics simulation model of a drug container-liquid-sterilization environment is constructed based on the finite element method, with the current physical state data of the container as input. Load the optimized process settings, simulate the entire sterilization process, and output key node data, including liquid stability, maximum minor deformation of the container, and fluctuation amplitude of the drug liquid. Calculate the reliability improvement of the optimized scheme. The calculation formula is: ； in, This represents the predicted score for the optimized sterilization effect. This represents the predicted sterilization effect score before optimization. When the improvement in reliability exceeds the reliability threshold, the final drug sterilization control plan is generated.

8. A real-time monitoring and optimization system for the sterilization process of pharmaceuticals, characterized in that, The system includes a data acquisition module, a container analysis module, a liquid internal analysis module, a liquid state determination module, a sterilization effect prediction module, an optimization module, and a control scheme generation module. The data acquisition module is used to collect multimodal monitoring data from the surface of a drug container in a drug sterilization environment. The multimodal monitoring data includes thermal imaging scan data and vibration wave signals. The container analysis module is used to acquire real-time physical state information of the drug container under high temperature and high pressure conditions, analyze the thermal imaging scan data, extract the geometric features of the drug container edge contour, calculate the degree of subtle deformation on the surface of the drug container, and determine the potential deformation location area. The liquid internal analysis module is used to perform detailed analysis on potential deformation locations. It processes the image data of potential deformation locations using a trained deep learning model to obtain abnormal fluctuation characteristics of the internal distribution of the drug liquid and determine the reaction state of the liquid during the sterilization process. The liquid state determination module is used to decompose the vibration fluctuation signal and match it with the abnormal fluctuation characteristics based on the abnormal fluctuation characteristics to determine the stability of the liquid state of the drug. The sterilization effect prediction module is used to generate a comprehensive influence vector based on the stability of the liquid state of the drug and the degree of minor deformation on the surface of the drug container. Key dimensions are extracted from the comprehensive influence vector to calculate the sterilization effect prediction score and quantify the drug quality assurance level. The optimization module is used to adjust the sterilization process parameters according to the comprehensive influence vector and generate optimized process settings; The control scheme generation module is used to input the current physical state information of the drug container through a multiphysics simulation model, verify the optimized process settings, judge the degree of reliability improvement of the sterilization process, and determine the final drug sterilization control scheme.

Images