Method for monitoring pharmaceutical production environment based on intelligent sensing system
By combining intelligent sensing systems with aerodynamic simulation models, environmental risk heat maps are generated and pollution source propagation paths are inverted, solving the problem of accurate source tracing of environmental anomalies in pharmaceutical production. This enables efficient environmental monitoring and material isolation, ensuring the safety and efficiency of pharmaceutical production.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- 河北省药品职业化检查员总队(南片区)(河北省疫苗检查中心)
- Filing Date
- 2026-04-30
- Publication Date
- 2026-07-24
Smart Images
Figure CN122448286A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of pharmaceutical production environment monitoring technology, and in particular to a method for pharmaceutical production environment monitoring based on an intelligent sensing system. Background Technology
[0002] In pharmaceutical manufacturing, the environmental parameters of the cleanroom directly affect drug quality. Current technologies typically deploy sensors for temperature, humidity, and particle counting within the cleanroom to collect and monitor these parameters. Some technologies generate environmental parameter distribution maps from the sensor data, enabling preliminary monitoring of the cleanroom environment. Simultaneously, pharmaceutical production requires batch production logs, detailing the operational steps, times, and corresponding areas for each process. However, existing monitoring technologies and production logs operate independently, failing to achieve data integration and correlation.
[0003] Current monitoring technologies can only collect environmental parameters and trigger threshold alarms, failing to correlate environmental anomalies with specific production processes. They struggle to distinguish which anomalies are associated with high-risk processes, resulting in a large amount of invalid data and an inability to accurately pinpoint the source of risk. Furthermore, when environmental anomalies occur, current technologies cannot trace the propagation path of pollution sources or the location of leaks, nor can they quantitatively analyze the decay patterns of pollutant concentrations, the duration of the anomaly, or its impact range. They can only conduct general environmental investigations and material isolation, which can easily lead to inaccurate isolation and untimely response, affecting pharmaceutical production efficiency and quality safety. Summary of the Invention
[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a method for monitoring the pharmaceutical production environment based on an intelligent sensing system.
[0005] To achieve the above objectives, the present invention adopts the following technical solution: a method for monitoring the pharmaceutical production environment based on an intelligent sensing system, comprising: An intelligent sensor array consisting of temperature and humidity sensors, micro differential pressure sensors, particle counters, and chemical gas sensors is deployed to continuously collect environmental parameters within the clean area and generate a preliminary environmental risk heat map. By introducing drug batch production logs, the preliminary environmental risk heat map is overlaid with specific process operation steps to filter out the spatiotemporal data subset that is strongly correlated with high-risk processes. For the aforementioned spatiotemporal data subset, the built-in aerodynamic simulation model is invoked to inversely calculate the propagation path of the pollution source and the coordinates of potential leakage points; Based on the coordinates of the leak point obtained by inversion calculation, the historical baseline data of the chemical gas sensor at the corresponding location is retrieved, and the concentration decay curve of the pollutants is calculated. The concentration decay curve was coupled with the air change rate data of the clean area to estimate the duration and radius of influence of this environmental anomaly. Based on the estimated radius of impact, the boundary of the contaminated material is delineated, and the corresponding environmental monitoring level upgrade instruction is triggered. By summarizing all intermediate data and decision-making instructions throughout the entire process, a structured environmental monitoring report is generated, which includes pollution source location, impact assessment, and disposal recommendations.
[0006] As a further aspect of the present invention, generating a preliminary environmental risk heat map includes: Continuous collection of environmental parameters within the clean area to form a raw environmental data stream; The original environmental data stream is time-stamped and mapped to coordinates, and the data points of the intelligent sensor array from different physical locations are uniformly converted into a spatiotemporal data matrix with spatial location labels. Based on the spatiotemporal data matrix, the dynamic environmental gradient of each monitoring point in the clean area is calculated. The dynamic environmental gradient reflects the physical trend of airflow organization and pollutant diffusion. The dynamic environmental gradient is compared with the preset drug manufacturing process environmental quality standards to identify abnormal fluctuation areas that deviate from the standards and generate a preliminary environmental risk heat map. The original environmental data stream is time-stamped and mapped to coordinates, transforming data points from the intelligent sensor array at different physical locations into a spatiotemporal data matrix with spatial location labels. Specifically, this includes: Read the factory calibration coefficient of each sensor in the intelligent sensor array, and perform unit normalization on the physical quantities in the original environmental data stream; The network time protocol is used to calibrate the clocks of all smart sensor arrays participating in data acquisition, eliminating time delay deviations generated during data transmission. Based on the pre-entered equipment layout drawings, each sensor is assigned a unique two-dimensional plane coordinate, which represents the physical installation location of the sensor in the production workshop. The normalized data is bound to the corresponding two-dimensional plane coordinates and calibrated timestamps to construct a three-dimensional tensor data structure. The three-dimensional tensor data structure is stacked according to the time series to form a continuous spatiotemporal data matrix. Each slice of the matrix represents a snapshot of the entire plant environment at a certain moment.
[0007] As a further aspect of the present invention, based on the spatiotemporal data matrix, the dynamic environmental gradient of each monitoring point within the clean area is calculated. This dynamic environmental gradient reflects the physical trends of airflow organization and pollutant diffusion, including: Extract continuous time series of specific monitoring indicators from the spatiotemporal data matrix, wherein the specific monitoring indicators include pressure difference and wind speed; The sliding window algorithm is used to calculate the difference between two adjacent time series data points to obtain the discrete spatial rate of change; Combining the physical distance between adjacent sensors, the discrete spatial rate of change is converted into an instantaneous change per unit distance, which is defined as a local gradient vector; By vector synthesis of local gradient vectors in all directions, the comprehensive dynamic environmental gradient of each monitoring point in each direction is calculated. The integrated dynamic environmental gradient is labeled at the corresponding position in the spatiotemporal data matrix to form gradient field distribution data.
[0008] As a further aspect of the present invention, the introduction of drug batch production logs, overlaying the preliminary environmental risk heat map with specific process operation steps, and filtering out a subset of spatiotemporal data strongly correlated with high-risk processes, includes: The production logs of the drug batches were analyzed to extract the start and end times of the feeding, liquid preparation, filling, sterilization and packaging processes, forming a process timeline; Map the process timeline onto the timeline of the preliminary environmental risk heatmap to align the time dimensions. Within each process time interval, the abnormal values of the corresponding area in the preliminary environmental risk heat map are integrated to obtain the cumulative environmental quality deviation value during the process. Set a deviation threshold, and determine the process time interval that exceeds the deviation threshold as a high-risk process interval; Extract all data blocks from the spatiotemporal data matrix corresponding to all high-risk process intervals to form the spatiotemporal data subset.
[0009] As a further aspect of the present invention, for the aforementioned spatiotemporal data subset, a built-in aerodynamic simulation model is invoked to inversely calculate the propagation path of the pollution source and the coordinates of potential leakage points, including: The abnormal concentration values in the aforementioned spatiotemporal data subset are used as input boundary conditions and imported into the aerodynamic simulation model. In the aerodynamic simulation model, a geometric model of the clean area building structure and equipment obstacles is set, and its surface roughness and adsorption coefficient are assigned. The aerodynamic simulation model is run to simulate the trajectory of pollutants in turbulent conditions and generate several pollution plume diffusion paths; For each of the pollution plume diffusion paths, reverse tracing is performed, and the starting point of the path is used as the coordinates of the candidate leakage point; Based on the fluid dynamic parameters of each path, the coordinates of the candidate leak points are weighted and scored, and the coordinate point with the highest score is selected as the final leak point coordinates.
[0010] As a further aspect of the present invention, the step of retrieving historical baseline data of the chemical gas sensor at the corresponding location based on the leak point coordinates obtained through inversion calculation, and calculating the concentration decay curve of the pollutants, includes: Based on the coordinates of the leak point, the nearest chemical gas sensor is matched in the equipment layout drawing to determine the target sensor identifier; Retrieve historical monitoring data of the target sensor identifier in the most recent production cycle from the long-term data storage area, calculate its average measurement value and standard deviation, and establish the historical baseline data; Obtain the concentration of the currently detected abnormal peak and calculate the offset of the abnormal peak concentration relative to the historical baseline data; Using the current moment as the zero point, and following the exponential decay law, the concentration decay curve is fitted using the offset. The concentration decay curve describes the theoretical process of pollutant concentration decreasing over time.
[0011] As a further aspect of the present invention, the concentration decay curve is coupled with the air change rate data of the clean area for analysis to estimate the duration and radius of influence of the current environmental anomaly, including: The current operating level of the mechanical ventilation system in the clean area is read from the production control system and converted into real-time air exchange rate data. Using the air exchange rate data, the complete air replacement cycle within the clean area is calculated; The complete replacement cycle is used as a time constant and substituted into the mathematical expression of the concentration decay curve to solve for the time span required for the concentration to return to the safe threshold, which is then determined as the duration. The maximum spatial diffusion distance of the pollutant is calculated by multiplying the average airflow velocity over the duration of the event, and this distance is determined as the radius of influence.
[0012] As a further aspect of the present invention, the step of delineating the contaminated material isolation boundary based on the estimated influence radius and triggering a corresponding environmental monitoring level upgrade instruction includes: Using the coordinates of the leak point as the center and the radius of influence as the radius, draw a circular warning area on the production workshop floor plan; The material storage racks, conveyor belts, and semi-finished product containers within the circular warning area are marked as objects to be isolated, and a material isolation instruction is generated. The material isolation instruction includes the geographical coordinate range of the circular warning area and a list of material codes. At the same time, the current environmental monitoring level parameters are changed from the normal mode to the enhanced mode, and the data sampling frequency of the intelligent sensor array is extended; Send the material isolation command and the modified monitoring configuration parameters to the production execution system.
[0013] As a further aspect of the present invention, the step of summarizing all intermediate data and decision instructions throughout the entire process to generate a structured environmental monitoring report containing pollution source location, impact assessment, and remediation recommendations includes: Create a report framework that predefines sections for pollution source description, impact scope, timeline, and treatment measures; Fill in the determined leak point coordinates and the calculated propagation path into the pollution source description section; Fill in the calculated radius of influence, duration of influence, and affected production batch number into the section on the scope of influence. Fill the timeline section with the results of the process overlay analysis and the time point that triggered the upgrade of the environmental monitoring level. Based on the chemical properties of the pollution source and the process it is located in, search the emergency response plan database, generate a targeted cleaning and disinfection procedure, and fill it into the treatment measures section. After filling in all fields in the report framework, export it as an encrypted electronic document.
[0014] As a further aspect of the present invention, the steps for constructing the aerodynamic simulation model include: Based on the architectural drawings of the clean area for pharmaceutical production, three-dimensional geometric information including walls, doors and windows, air supply outlets, return air outlets, and the outer contour of production equipment is extracted. The extracted three-dimensional geometric information is imported into computer-aided design software to establish a three-dimensional solid computational domain mesh model of the clean area; Define physical boundary conditions for different surfaces in the three-dimensional solid computational domain mesh model. The physical boundary conditions include setting the air supply vent as a constant velocity inlet boundary condition, setting the return air vent as a free outflow boundary condition, and setting the wall surface and equipment surface as a no-slip wall boundary condition. In the three-dimensional solid computational domain mesh model, the physical property parameters of air are set, including air density, dynamic viscosity, specific heat capacity and thermal conductivity. The Navier-Stokes equations based on Reynolds averages are selected as the core governing equations of the model, and combined with the standard turbulence model to describe the air flow state in the clean area. The control equations are discretized, and the finite volume method is used to solve them numerically on the computational domain grid model, thus constructing an aerodynamic simulation model that can simulate the diffusion process of pollutants under turbulent conditions.
[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: This solution overlays a preliminary environmental risk heatmap generated from environmental parameters collected by an intelligent sensor array onto specific process steps in the drug batch production log, selecting a subset of spatiotemporal data strongly correlated with high-risk processes. This approach breaks down the barriers between environmental parameter monitoring and drug production processes, ensuring that environmental monitoring data is no longer isolated parameter records. It enables precise filtering of environmental data related to high-risk processes, eliminating irrelevant and invalid monitoring information. This makes the investigation of environmental anomalies more targeted, avoiding the waste of manpower and time caused by blind investigations, and making environmental monitoring more aligned with the actual needs of drug production, achieving precise focus on environmental risks.
[0016] For the selected spatiotemporal data subset, the built-in aerodynamic simulation model is used to inversely calculate the propagation path of the pollution source and the coordinates of potential leak points. Then, based on the leak point coordinates, historical baseline data from the corresponding chemical gas sensors are retrieved to calculate the concentration decay curve of the pollutants. This concentration decay curve is then coupled with the air exchange rate data of the clean area to estimate the duration and radius of influence of the environmental anomaly. This solution enables precise source tracing and quantitative analysis of environmental anomalies, clearly identifying the specific location and propagation patterns of the pollution source, and clearly understanding the scope and duration of the environmental anomaly's impact. It eliminates the need for comprehensive, blind investigations and large-scale material isolation, accurately delineating the isolation boundaries of contaminated materials, reducing unnecessary material losses. Furthermore, by precisely triggering environmental monitoring level upgrade commands, it makes environmental anomaly handling more scientific, avoiding problems of untimely or excessive handling, and ensuring the continuity and safety of pharmaceutical production. Attached Figure Description
[0017] Figure 1 This is a flowchart of the drug production environment monitoring method based on an intelligent sensing system according to the present invention; Figure 2 A flowchart for dynamic environment gradient calculation; Figure 3 A flowchart for filtering spatiotemporal data subsets of high-risk processes; Figure 4 This is a graph showing the correlation between the duration of environmental anomalies and their radius of influence. Figure 5 This is a graph showing the exponential decay of pollutant concentration. Detailed Implementation
[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.
[0019] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.
[0020] See Figure 1 An intelligent sensor array, consisting of temperature and humidity sensors, micro-differential pressure sensors, particle counters, and chemical gas sensors, is deployed within the cleanroom of pharmaceutical production. This array continuously collects environmental parameters within the cleanroom, and the acquired data, after preliminary processing, generates a preliminary environmental risk heat map reflecting abnormal fluctuations. By incorporating pharmaceutical batch production logs, the preliminary environmental risk heat map is overlaid with specific process operation steps on a timeline for analysis, thereby identifying a subset of spatiotemporal data strongly correlated with high-risk processes. For this subset of spatiotemporal data, the system invokes a built-in aerodynamic simulation model, using the abnormal data as input, to inversely calculate the propagation path of pollutants and the coordinates of potential leak points. Based on the leak point coordinates obtained from the inverse calculation, the system retrieves historical baseline data from the corresponding chemical gas sensors and calculates the current pollutant concentration decay curve through comparative analysis. This concentration decay curve is then coupled with real-time air exchange rate data of the cleanroom read from the production control system for analysis, thereby estimating the duration and radius of influence of this environmental anomaly. Based on the estimated impact radius, the system delineates a circular warning zone on the production workshop floor plan as the boundary for isolating contaminated materials. It then sends an isolation command containing the zone's coordinates and a bill of materials to the production execution system. Simultaneously, it triggers an environmental monitoring level upgrade command, adjusting the sampling frequency of the sensor array. The system aggregates all intermediate data and decision commands generated throughout the process, automatically populating the data according to a pre-defined report framework to generate a structured environmental monitoring report that includes pollution source location, impact assessment, and remediation recommendations.
[0021] In one embodiment of the present invention, an intelligent sensor array deployed in a cleanroom for pharmaceutical production operates continuously. Temperature and humidity sensors, micro-differential pressure sensors, particle counters, and chemical gas sensors collect corresponding parameters, forming a raw environmental data stream containing multidimensional physical quantities. The raw environmental data stream is then time-stamped and mapped to coordinates. This process begins by reading the factory calibration coefficients of each sensor in the intelligent sensor array, normalizing the physical quantities in the raw environmental data stream, and converting voltage or current signals into unified SI values. A network time protocol is used to calibrate the clocks of all intelligent sensor arrays participating in data acquisition, eliminating time delays caused by differences in network transmission and device response, ensuring all data points have a unified time reference. Based on pre-recorded cleanroom equipment layout drawings, each sensor is assigned a unique two-dimensional plane coordinate, which precisely corresponds to the sensor's physical installation location in the production workshop. The normalized data is bound to the corresponding two-dimensional plane coordinate and the calibrated timestamp to construct a three-dimensional tensor data structure. The three dimensions of this structure correspond to the time series, sensor spatial location index, and monitoring indicator type, respectively. The three-dimensional tensor data structure is stacked sequentially according to the time series to form a continuous spatiotemporal data matrix. Each time slice of the spatiotemporal data matrix represents a snapshot of the entire plant environment at a specific moment.
[0022] In some embodiments, the dynamic environmental gradient of each monitoring point within the clean area is calculated based on a spatiotemporal data matrix. This dynamic environmental gradient reflects the physical trends of airflow organization and pollutant diffusion. See also... Figure 2 Continuous time series of specific monitoring indicators are extracted from the spatiotemporal data matrix. These indicators explicitly include pressure difference and wind speed. A sliding window algorithm with a fixed time length is used to calculate the difference between two adjacent time series data points, yielding a discrete spatial rate of change. Combined with the actual physical distance between adjacent sensors, the discrete spatial rate of change is converted into an instantaneous change per unit distance, defined as a local gradient vector. In two-dimensional plane coordinates, for any monitoring point, its local gradient vectors in the east-west and north-south directions need to be calculated. The local gradient vectors in all directions are then vector-synthesized using the following formula:
[0023] in: Indicates at time ,coordinate The magnitude of the gradient vector of the integrated dynamic environment at that location. This represents the physical quantity being monitored (such as pressure difference or wind speed). and These represent the rate of change per unit distance in the east-west and north-south directions, respectively, i.e., the components of the local gradient vector. After calculating the comprehensive dynamic environmental gradient of each monitoring point in each direction, the comprehensive dynamic environmental gradient is marked on the corresponding coordinate position of the spatiotemporal data matrix, forming gradient field distribution data covering the entire clean area. Finally, the dynamic environmental gradient reflected in the gradient field distribution data is compared with the preset environmental quality standards for pharmaceutical manufacturing processes to identify areas with abnormal gradients or values deviating from the standards, generating a preliminary environmental risk heat map that uses different colors or numerical intensities to indicate the degree of abnormality.
[0024] In one embodiment of the present invention, see [reference] Figure 3 The process involves analyzing drug batch production logs to extract the start and end times of each stage: material feeding, liquid preparation, filling, sterilization, and packaging, forming a process timeline. This timeline is then mapped onto the timeline of a preliminary environmental risk heatmap to align the time dimensions. Within each process time interval, the abnormal values in the corresponding areas of the preliminary environmental risk heatmap are integrated to obtain the cumulative environmental quality deviation value for that process. A deviation threshold is set, and process time intervals exceeding this threshold are classified as high-risk process intervals. Data blocks from the spatiotemporal data matrix corresponding to all high-risk process intervals are extracted to form a spatiotemporal data subset strongly correlated with the high-risk processes.
[0025] In practice, the pharmaceutical batch production log is introduced into the system as a structured production record. The log clearly records each operational step, including material feeding, solution preparation, filling, sterilization, and packaging, along with its precise timestamps. The pharmaceutical batch production log is analyzed to extract the start and end times of each operational step. These time points are arranged sequentially to form a complete timeline covering the entire process of a single production batch. This timeline is then mapped onto the timeline of a preliminary environmental risk heatmap, ensuring precise alignment between each time interval of the timeline and the preliminary environmental risk heatmap in the time dimension. This guarantees that each operational step can be found in the spatiotemporal representation of the heatmap.
[0026] In some embodiments, within each process time interval defined by the process time axis, the abnormal values of the corresponding physical regions in the preliminary environmental risk heatmap are integrated. The purpose of the integration is to quantify the overall deviation of environmental risk during that process. The calculation formula is:
[0027] in: This represents the cumulative deviation of environmental quality calculated within a specific process time interval. and These represent the start and end times of the process, respectively. This represents the spatial extent of the main production area involved in this process in a two-dimensional plane coordinate system. This is a preliminary environmental risk heat map on the coordinate system. and time The defined risk value function. The result of this operation is a scalar value characterizing the cumulative deviation of environmental quality during the process.
[0028] Optionally, a preset deviation threshold is set, and the calculated cumulative environmental quality deviation value is compared with the deviation threshold. The system identifies process time intervals exceeding the deviation threshold as high-risk process intervals, meaning that significant and continuous anomalies occurred in the environmental parameters of the clean area during process operations within that time period. Data blocks are extracted from the spatiotemporal data matrix corresponding to all high-risk process intervals. These data blocks contain raw and derived data from all sensors collected from the high-risk spatial area within the high-risk process intervals. These data blocks together form a spatiotemporal data subset strongly correlated with the high-risk processes.
[0029] In one embodiment of the present invention, for a selected subset of spatiotemporal data, the system invokes a built-in aerodynamic simulation model to perform pollution source inversion calculation. Abnormal concentration values within the spatiotemporal data subset are imported into the aerodynamic simulation model as input boundary conditions. In the aerodynamic simulation model, a three-dimensional geometric model of the clean area building structure and equipment obstacles is first set, and specific physical parameters, including surface roughness and adsorption coefficient, are assigned to the surfaces of these geometric models. The aerodynamic simulation model is run to simulate the trajectory of pollutants in turbulent flow under set supply and return air conditions, calculating and generating several possible pollution plume diffusion paths. Each pollution plume diffusion path calculated by the model is traced in reverse, that is, the source is traced back along the simulated flow direction, and the starting position of the path is used as the coordinates of a candidate leak point.
[0030] In some embodiments, the coordinates of candidate leak points are weighted and scored based on the hydrodynamic parameters of each contaminant plume diffusion path to determine the most likely leak location. The calculation formula is as follows:
[0031] in: The final score represents the coordinates of a candidate leak point. It represents the number of pollution plume diffusion paths traced back to that coordinate. It is the first The confidence parameter for each path can be calculated based on the goodness of fit between the path and the sensor's measured data. It corresponds to the first The weighting coefficients for each path are determined by the path's flow energy or initial release intensity. The scores of all calculated candidate leak point coordinates are compared, and the coordinates with the highest score are selected as the final leak point coordinates determined by the system.
[0032] Optionally, the aerodynamic simulation model is constructed following standard computational fluid dynamics procedures. Based on the architectural drawings of the pharmaceutical production clean area, all three-dimensional geometric information, including walls, doors and windows, air supply vents, return air vents, and the outer contours of production equipment, is extracted. The extracted three-dimensional geometric information is imported into computer-aided design software for geometric cleanup and repair, and meshing is performed to establish a three-dimensional solid computational domain mesh model of the clean area. Physical boundary conditions are defined for different surfaces in the three-dimensional solid computational domain mesh model. These physical boundary conditions include setting the air supply vents as constant velocity inlet boundary conditions, the return air vents as free outflow boundary conditions, and the wall and equipment surfaces as no-slip wall boundary conditions.
[0033] It is understandable that the physical properties of air need to be defined in the three-dimensional solid computational domain mesh model. These physical properties include air density, dynamic viscosity, specific heat capacity, and thermal conductivity. The Navier-Stokes equations based on Reynolds averages are chosen as the core governing equations of the aerodynamic simulation model, combined with a standard turbulence model to describe the airflow state within the clean area. The governing equations are discretized, and the finite volume method is used for numerical solution on the computational domain mesh model. Stable solutions to the flow field are obtained through iterative calculations, thereby constructing an aerodynamic simulation model capable of simulating the diffusion process of pollutants under turbulent conditions.
[0034] In one embodiment of the present invention, based on the coordinates of the leak point obtained through inversion calculation, the nearest chemical gas sensor is matched in the equipment layout drawing to determine the target sensor identifier. Historical monitoring data of the target sensor identifier within the most recent production cycle is retrieved from the long-term data storage area, and its average measurement value and standard deviation are calculated to establish historical baseline data. The currently detected abnormal peak concentration is obtained, and the offset of the abnormal peak concentration relative to the historical baseline data is calculated. Taking the current moment as the zero point, according to the exponential decay law, the concentration decay curve of the pollutant is fitted using the offset, which describes the theoretical process of the pollutant concentration decreasing over time.
[0035] The current operating level of the mechanical ventilation system in the clean area is read from the production control system and converted into real-time air exchange rate data. Using this data, the complete air replacement cycle within the clean area is calculated. This cycle is then used as a time constant and substituted into the mathematical expression for the concentration decay curve to determine the time span required for the concentration to return to the safe threshold, which is defined as the duration of this environmental anomaly. Based on the average airflow velocity over this duration, multiplied by the duration, the maximum spatial diffusion distance of the pollutants is calculated, and this distance is determined as the radius of influence of this environmental anomaly.
[0036] In some embodiments, the current operating level of the mechanical ventilation system in the clean area is directly read from the control system of the pharmaceutical production environment. There is a definite correspondence between the operating level and the air exchange rate in the clean area. See Table 1 for an example of the correspondence between the ventilation system operating level and the air exchange rate.
[0037] Table 1: Correspondence between Ventilation System Operating Levels and Air Change Rates
[0038] Based on Table 1, the read operating speed is converted into real-time air exchange rate data. Using the air exchange rate data, the time required for the air in the clean area to be completely replaced once, i.e., the complete replacement cycle, is calculated. The calculated complete replacement cycle is used as a time constant and substituted into the mathematical expression of the concentration decay curve to solve for the time span required for the pollutant concentration to decrease from the current abnormal peak to the preset safety threshold. This time span is determined as the duration of this environmental anomaly.
[0039] When estimating the radius of influence of pollutants, the assumption of uniform linear diffusion is abandoned, and spatial interpolation calculations are performed based on the three-dimensional concentration field data output by the constructed aerodynamic simulation model. The moment when the concentration returns to the safe threshold in the concentration decay curve is used as the time boundary condition and imported into the aerodynamic simulation model that has completed convergence calculations. In the model, the coordinates of the leak point obtained by inversion are used as the spatial origin, and the position of the pollutant concentration front at that moment is extracted along multiple dominant directions of the actual airflow organization in the clean area, including the mainstream direction from the air supply outlet to the return air outlet and the flow around the equipment caused by obstacles. The Euclidean distance between the front position in each direction and the origin is fitted to form a closed isoconcentration line. The maximum geometric span of the area covered by this isoconcentration line is defined as the actual radius of influence of this environmental anomaly. This process directly utilizes the turbulent field and obstacle flow around effect in the simulation model, avoiding the fundamental errors caused by the assumption of uniform linear motion, and ensuring that the calculation result of the radius of influence is strictly limited by the physical boundary and airflow organization characteristics of the clean room, usually on the meter scale.
[0040] The current operating level of the mechanical ventilation system in the clean area is read from the production control system and converted into real-time air exchange rate data. Using this data, the complete air replacement cycle within the clean area is calculated. This cycle is then used as a time constant and substituted into the mathematical expression for the concentration decay curve to determine the time span required for the concentration to return to the safe threshold, which is defined as the duration of this environmental anomaly. Based on the average airflow velocity over this duration, combined with the spatial distribution of the concentration field output from the aerodynamic simulation model, the maximum spatial diffusion distance of the pollutants is calculated and determined as the radius of influence of this environmental anomaly.
[0041] See Figure 4 This is a correlation chart analyzing the duration of environmental anomalies and their impact radius, showing the relationship between airflow diffusion characteristics and the impact radius of environmental anomalies under different ventilation levels. Higher ventilation levels result in shorter durations of environmental anomalies. This chart helps production managers quickly assess the potential spatial impact range of environmental anomalies under different ventilation conditions and airflow velocities, providing data support for delineating material isolation boundaries. When an environmental anomaly occurs, increasing the ventilation level can effectively shorten the duration of the anomaly, thereby significantly reducing the impact radius and minimizing the impact on adjacent production areas and materials. The analysis results can serve as an important basis for developing emergency response plans for environmental anomalies, used to delineate circular warning zones and material isolation boundaries under different circumstances.
[0042] In this method, the computational domain mesh model of the aerodynamic simulation model is strictly established according to the actual architectural drawings of the clean area for pharmaceutical production, and its three-dimensional geometric coordinate range precisely corresponds to the physical dimensions of the production workshop. When estimating the radius of influence, the concentration front distance extracted from the model is the actual physical distance under this geometric coordinate system. The system has built-in scale verification logic: if the radius of influence calculated by the model exceeds the physical boundary of the workshop or the preset maximum reasonable threshold, it is automatically clamped to the limit distance from the leak point to the farthest return air vent within the workshop. In this way, it is ensured that the estimated result of the radius of influence is always within the actual spatial scale of the cleanroom, avoiding invalid values that far exceed the workshop dimensions due to mathematical model simplification, and ensuring the feasibility of the technical solution on the pharmaceutical production site.
[0043] In one embodiment of the invention, a circular warning zone is drawn on the production workshop floor plan, centered on the coordinates of the leak point and with the calculated radius of influence as the radius. Material storage racks, conveyor belts, and semi-finished product containers within the circular warning zone are marked as objects to be isolated, generating a material isolation instruction. This instruction includes the geographical coordinates of the circular warning zone and a list of material codes. Simultaneously, the current environmental monitoring level parameters are changed from the normal mode to the enhanced mode, and the data sampling frequency of the intelligent sensor array is extended. The material isolation instruction and the modified monitoring configuration parameters are then sent to the production execution system.
[0044] Create a report framework with predefined sections for pollution source description, impact range, timeline, and treatment measures. Enter the determined leak point coordinates and estimated propagation path into the pollution source description section. Enter the calculated impact radius, duration, and affected production batch numbers into the impact range section. Enter the results of the process overlay analysis and the time point triggering the upgrade of the environmental monitoring level into the timeline section. Based on the chemical properties of the pollution source and its associated process, search the emergency response plan database to generate targeted cleaning and disinfection procedures, and enter these into the treatment measures section. After filling in all fields in the report framework, export it as an encrypted electronic document.
[0045] In practice, using the leak point coordinates determined in the aforementioned steps as the geometric center and the calculated radius of influence as the length, a circular warning zone is drawn on the digital floor plan of the production workshop. All material storage facilities, conveying devices, and semi-finished product containers located within this circular warning zone are highlighted on the floor plan; these marked objects are identified by the system as objects to be isolated. Based on the identification results, a structured material isolation instruction is generated, containing the precise geographical coordinate boundary information of the circular warning zone and a unique code list of all materials to be isolated within the zone. Simultaneously, the system modifies the current environmental monitoring level parameters from the normal monitoring mode to the enhanced monitoring mode. This modification includes extending the data sampling frequency of all sensors in the intelligent sensor array, for example, from once per minute to once every ten seconds. After completing the above instruction generation and parameter modification, the system synchronously sends the material isolation instruction and new monitoring configuration parameters to the pharmaceutical production execution system to drive physical isolation at the production site and enhance monitoring intensity.
[0046] In some embodiments, the system creates a report framework with a standardized chapter structure. This framework predefines chapters for pollution source description, impact range, timeline, and treatment measures. The pollution source description chapter is filled with the coordinates of the final leak point obtained from the aerodynamic simulation model inversion and the simulated main pollution plume diffusion path. The impact range chapter is filled with the calculated specific impact radius, the estimated duration of the environmental anomaly, and the relevant production batch numbers within the impact radius. The timeline chapter is filled with the high-risk process intervals identified in the process overlay analysis, the time point when the environmental anomaly was first detected by the sensors, and the time point when the system automatically triggered the environmental monitoring level upgrade command.
[0047] Optionally, the system retrieves an integrated emergency response plan database based on the type of chemical substance identified in the pollution source and the specific production process where the leak point is located. This database stores standardized handling procedures for different chemicals in different production areas. The system matches and generates a targeted cleaning and disinfection procedure based on the search results, and fills this procedure text into the handling measures section of the report framework. After all predefined fields in the report framework are filled in, the system encapsulates the complete report content and uses an encryption algorithm to generate an encrypted electronic document for download or transmission.
[0048] Understandably, the delineation of circular warning zones transforms the abstract radius of influence into a concrete geographic fence; material isolation instructions are directly linked to environmental anomalies and material quality status; and instructions to upgrade environmental monitoring levels enable the dynamic allocation of monitoring resources. The automatic generation of structured environmental monitoring reports summarizes the entire process of data and decision-making logic, from anomaly detection and source analysis to impact assessment. The pollution source description section provides location information for pollution events, the impact range section quantifies the consequences of the event, the timeline section records the event evolution and system response process, and the handling measures section provides specific action guidelines.
[0049] See Figure 5 This is an exponential decay curve of pollutant concentration, illustrating the exponential decay process of pollutant concentration over time in pharmaceutical production environment monitoring. By fitting the curve, the time required for the concentration to return to the safe threshold can be accurately calculated, i.e., the duration of the environmental anomaly. The time constant τ is determined by the air exchange rate of the ventilation system; increasing the ventilation level reduces τ, thereby accelerating concentration decay and shortening the duration of the anomaly. The curve allows for precise calculation of the time required for the concentration to return to the safe threshold, i.e., the duration of the environmental anomaly—core data for delineating isolation boundaries and developing response plans. Different ventilation levels correspond to different time constants (τ). By comparing different curves, the significant effect of increasing the ventilation level on shortening the duration of the anomaly can be visually demonstrated, providing quantitative support for emergency decision-making.
[0050] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.
Claims
1. A method for monitoring the pharmaceutical production environment based on an intelligent sensing system, characterized in that, include: An intelligent sensor array consisting of temperature and humidity sensors, micro differential pressure sensors, particle counters, and chemical gas sensors is deployed to continuously collect environmental parameters within the clean area and generate a preliminary environmental risk heat map. By introducing drug batch production logs, the preliminary environmental risk heat map is overlaid with specific process operation steps to filter out the spatiotemporal data subset that is strongly correlated with high-risk processes. For the aforementioned spatiotemporal data subset, the built-in aerodynamic simulation model is invoked to inversely calculate the propagation path of the pollution source and the coordinates of potential leakage points; Based on the coordinates of the leak point obtained by inversion calculation, the historical baseline data of the chemical gas sensor at the corresponding location is retrieved, and the concentration decay curve of the pollutants is calculated. The concentration decay curve was coupled with the air change rate data of the clean area to estimate the duration and radius of influence of this environmental anomaly. Based on the estimated radius of impact, the boundary of the contaminated material is delineated, and the corresponding environmental monitoring level upgrade instruction is triggered. By summarizing all intermediate data and decision-making instructions throughout the entire process, a structured environmental monitoring report is generated, which includes pollution source location, impact assessment, and disposal recommendations.
2. The method for monitoring the pharmaceutical production environment based on an intelligent sensing system as described in claim 1, characterized in that, Generate a preliminary environmental risk heat map, including: Continuous collection of environmental parameters within the clean area to form a raw environmental data stream; The original environmental data stream is time-stamped and mapped to coordinates, and the data points of the intelligent sensor array from different physical locations are uniformly converted into a spatiotemporal data matrix with spatial location labels. Based on the spatiotemporal data matrix, the dynamic environmental gradient of each monitoring point in the clean area is calculated. The dynamic environmental gradient reflects the physical trend of airflow organization and pollutant diffusion. The dynamic environmental gradient is compared with the preset drug manufacturing process environmental quality standards to identify abnormal fluctuation areas that deviate from the standards and generate a preliminary environmental risk heat map. The original environmental data stream is time-stamped and mapped to coordinates, transforming data points from the intelligent sensor array at different physical locations into a spatiotemporal data matrix with spatial location labels. Specifically, this includes: Read the factory calibration coefficient of each sensor in the intelligent sensor array, and perform unit normalization on the physical quantities in the original environmental data stream; The network time protocol is used to calibrate the clocks of all smart sensor arrays participating in data acquisition, eliminating time delay deviations generated during data transmission. Based on the pre-entered equipment layout drawings, each sensor is assigned a unique two-dimensional plane coordinate, which represents the physical installation location of the sensor in the production workshop. The normalized data is bound to the corresponding two-dimensional plane coordinates and calibrated timestamps to construct a three-dimensional tensor data structure. The three-dimensional tensor data structure is stacked according to the time series to form a continuous spatiotemporal data matrix. Each slice of the matrix represents a snapshot of the entire plant environment at a certain moment.
3. The method for monitoring the pharmaceutical production environment based on an intelligent sensing system as described in claim 2, characterized in that, Based on the aforementioned spatiotemporal data matrix, the dynamic environmental gradient at each monitoring point within the clean area is calculated. This dynamic environmental gradient reflects the physical trends of airflow organization and pollutant diffusion, including: Extract continuous time series of specific monitoring indicators from the spatiotemporal data matrix, wherein the specific monitoring indicators include pressure difference and wind speed; The sliding window algorithm is used to calculate the difference between two adjacent time series data points to obtain the discrete spatial rate of change; Combining the physical distance between adjacent sensors, the discrete spatial rate of change is converted into an instantaneous change per unit distance, which is defined as a local gradient vector; By vector synthesis of local gradient vectors in all directions, the comprehensive dynamic environmental gradient of each monitoring point in each direction is calculated. The integrated dynamic environmental gradient is labeled at the corresponding position in the spatiotemporal data matrix to form gradient field distribution data.
4. The method for monitoring the pharmaceutical production environment based on an intelligent sensing system as described in claim 3, characterized in that, The introduction of drug batch production logs involves overlaying the preliminary environmental risk heatmap with specific process operation steps to filter out a spatiotemporal data subset strongly correlated with high-risk processes, including: The production logs of the drug batches were analyzed to extract the start and end times of the feeding, liquid preparation, filling, sterilization and packaging processes, forming a process timeline; Map the process timeline onto the timeline of the preliminary environmental risk heatmap to align the time dimensions. Within each process time interval, the abnormal values of the corresponding area in the preliminary environmental risk heat map are integrated to obtain the cumulative environmental quality deviation value during the process. Set a deviation threshold, and determine the process time interval that exceeds the deviation threshold as a high-risk process interval; Extract all data blocks from the spatiotemporal data matrix corresponding to all high-risk process intervals to form the spatiotemporal data subset.
5. The method for monitoring the pharmaceutical production environment based on an intelligent sensing system as described in claim 4, characterized in that, For the aforementioned spatiotemporal data subset, the built-in aerodynamic simulation model is invoked to inversely calculate the propagation path of the pollution source and the coordinates of potential leakage points, including: The abnormal concentration values in the aforementioned spatiotemporal data subset are used as input boundary conditions and imported into the aerodynamic simulation model. In the aerodynamic simulation model, a geometric model of the clean area building structure and equipment obstacles is set, and its surface roughness and adsorption coefficient are assigned. The aerodynamic simulation model is run to simulate the trajectory of pollutants in turbulent conditions and generate several pollution plume diffusion paths; For each of the pollution plume diffusion paths, reverse tracing is performed, and the starting point of the path is used as the coordinates of the candidate leakage point; Based on the fluid dynamic parameters of each path, the coordinates of the candidate leak points are weighted and scored, and the coordinate point with the highest score is selected as the final leak point coordinates.
6. The method for monitoring the pharmaceutical production environment based on an intelligent sensing system as described in claim 5, characterized in that, The process involves retrieving historical baseline data from the chemical gas sensor at the corresponding location based on the leak point coordinates obtained through inversion calculation, and calculating the pollutant concentration decay curve, including: Based on the coordinates of the leak point, the nearest chemical gas sensor is matched in the equipment layout drawing to determine the target sensor identifier; Retrieve historical monitoring data of the target sensor identifier in the most recent production cycle from the long-term data storage area, calculate its average measurement value and standard deviation, and establish the historical baseline data; Obtain the concentration of the currently detected abnormal peak and calculate the offset of the abnormal peak concentration relative to the historical baseline data; Using the current moment as the zero point, and following the exponential decay law, the concentration decay curve is fitted using the offset. The concentration decay curve describes the theoretical process of pollutant concentration decreasing over time.
7. The method for monitoring the pharmaceutical production environment based on an intelligent sensing system as described in claim 6, characterized in that, The concentration decay curve was coupled with the air change rate data of the clean area for analysis to estimate the duration and radius of influence of this environmental anomaly, including: The current operating level of the mechanical ventilation system in the clean area is read from the production control system and converted into real-time air exchange rate data. Using the air exchange rate data, the complete air replacement cycle within the clean area is calculated; The complete replacement cycle is used as a time constant and substituted into the mathematical expression of the concentration decay curve to solve for the time span required for the concentration to return to the safe threshold, which is then determined as the duration. The maximum spatial diffusion distance of the pollutant is calculated by multiplying the average airflow velocity over the duration of the event, and this distance is determined as the radius of influence.
8. The method for monitoring the pharmaceutical production environment based on an intelligent sensing system as described in claim 7, characterized in that, The process of defining the contaminated material isolation boundary based on the estimated radius of influence and triggering a corresponding environmental monitoring level upgrade instruction includes: Using the coordinates of the leak point as the center and the radius of influence as the radius, draw a circular warning area on the production workshop floor plan; The material storage racks, conveyor belts, and semi-finished product containers within the circular warning area are marked as objects to be isolated, and a material isolation instruction is generated. The material isolation instruction includes the geographical coordinate range of the circular warning area and a list of material codes. At the same time, the current environmental monitoring level parameters are changed from the normal mode to the enhanced mode, and the data sampling frequency of the intelligent sensor array is extended; Send the material isolation command and the modified monitoring configuration parameters to the production execution system.
9. The method for monitoring the pharmaceutical production environment based on an intelligent sensing system as described in claim 8, characterized in that, The process aggregates all intermediate data and decision-making instructions throughout the entire process to generate a structured environmental monitoring report that includes pollution source location, impact assessment, and remediation recommendations, including: Create a report framework that predefines sections for pollution source description, impact scope, timeline, and treatment measures; Fill in the determined leak point coordinates and the calculated propagation path into the pollution source description section; Fill in the calculated radius of influence, duration of influence, and affected production batch number into the section on the scope of influence. Fill the timeline section with the results of the process overlay analysis and the time point that triggered the upgrade of the environmental monitoring level. Based on the chemical properties of the pollution source and the process it is located in, search the emergency response plan database, generate a targeted cleaning and disinfection procedure, and fill it into the treatment measures section. After filling in all fields in the report framework, export it as an encrypted electronic document.
10. The method for monitoring the pharmaceutical production environment based on an intelligent sensing system as described in claim 9, characterized in that, The steps for constructing the aerodynamic simulation model include: Based on the architectural drawings of the clean area for pharmaceutical production, three-dimensional geometric information including walls, doors and windows, air supply outlets, return air outlets, and the outer contour of production equipment is extracted. The extracted three-dimensional geometric information is imported into computer-aided design software to establish a three-dimensional solid computational domain mesh model of the clean area; Define physical boundary conditions for different surfaces in the three-dimensional solid computational domain mesh model. The physical boundary conditions include setting the air supply vent as a constant velocity inlet boundary condition, setting the return air vent as a free outflow boundary condition, and setting the wall and equipment surfaces as non-slip wall boundary conditions. In the three-dimensional solid computational domain mesh model, the physical property parameters of air are set, including air density, dynamic viscosity, specific heat capacity and thermal conductivity. The Navier-Stokes equations based on Reynolds averages are selected as the core governing equations of the model, and combined with the standard turbulence model to describe the air flow state in the clean area. The control equations are discretized, and the finite volume method is used to solve them numerically on the computational domain grid model, thus constructing an aerodynamic simulation model that can simulate the diffusion process of pollutants under turbulent conditions.