Air quality monitoring method and device based on medical clean space and related equipment

Abstract

The invention relates to the technical field of artificial intelligence and automatic environment monitoring, discloses an air quality monitoring method, device and equipment based on a medical clean space and a medium, and aims to solve the problem of low automatic air quality monitoring efficiency of the medical clean space in the prior art. When it is judged that the air quality of a target area of the medical clean space is abnormal, an early warning signal is generated; taking the early warning signal as a task triggering instruction, and driving a target mobile robot; receiving returned data from the target mobile robot, the returned data being verification data generated after the target mobile robot responds to the task triggering instruction, autonomously plans a path and moves to the target area based on the position information contained in the early warning signal, and executes environment data collection and field condition verification; and on the basis of the verification data, the early warning signal is confirmed or released, so that the automatic monitoring efficiency of the air quality is improved by realizing an automatic monitoring mechanism of'prediction-driving-verification '.

Description

Technical Field

[0001] This invention relates to the fields of artificial intelligence and automatic environmental monitoring technology, and in particular to a method, device, equipment and medium for air quality monitoring in medical clean spaces. Background Technology

[0002] Medical clean spaces refer to places with requirements for air cleanliness and quality, such as operating rooms, sterile preparation workshops, and intensive care units (ICUs). In traditional technologies, air quality monitoring in medical clean spaces mainly relies on static monitoring schemes based on fixed sensor networks. This scheme typically includes: deploying fixed environmental sensors at key points in the clean space (such as air vents and around the operating table) to continuously collect data such as suspended particles, temperature, humidity, and pressure differentials; when the reading of any sensor exceeds a preset safety threshold, the system triggers an audible and visual alarm, notifying management personnel to conduct manual intervention and investigation.

[0003] However, after in-depth analysis, the inventors discovered two inherent technical flaws in this traditional solution, resulting in low monitoring efficiency and delayed response: First, the static and limited nature of the monitoring means that the deployment location and monitoring range of the fixed sensors are pre-set and static, failing to cover all areas of the clean space. Furthermore, it cannot effectively track localized and instantaneous pollution caused by dynamic events such as personnel movement and equipment movement. This leads to a one-sided and delayed perception of air pollution events, making it impossible to achieve dynamic and timely monitoring of the air quality of the entire space. Second, in traditional technology, the risk is only "informed" after an air quality alarm is triggered, while subsequent critical steps such as locating the pollution source and verifying abnormal conditions rely entirely on experienced personnel for on-site handling. This manual intervention response mode is not only inefficient but also prolongs the risk exposure time. In summary, the core problem with traditional technologies lies in the combination of "static and limited monitoring methods" and "reliance on manual response mechanisms," which leads to incomplete and inaccurate perception of air quality risks in medical clean spaces and inefficient response and handling. This means that traditional technologies can only achieve passive, post-event alarms and cannot achieve proactive early warning and rapid closed-loop handling of potential risks.

[0004] Therefore, improving the intelligence and efficiency of traditional technologies for automatic air quality monitoring in medical clean spaces has become an urgent technical problem to be solved in air quality monitoring in medical clean spaces such as operating rooms, sterile preparation workshops, and ICU wards. Summary of the Invention

[0005] This invention provides a method, device, computer equipment, and medium for air quality monitoring in medical clean spaces, in order to solve the technical problem of low efficiency in automatic air quality monitoring in medical clean spaces in traditional technologies.

[0006] In a first aspect, an air quality monitoring method based on medical clean spaces is provided, comprising: generating and outputting a corresponding early warning signal when a predictive model determines that an air quality anomaly will occur in a target area of ​​the medical clean space in the future; using the early warning signal as a task triggering command to drive a target mobile robot; receiving feedback data from the target mobile robot, wherein the feedback data is verification data generated by the target mobile robot in response to the task triggering command, autonomously planning a path and moving to the target area based on the location information contained in the early warning signal, and performing environmental data collection and on-site condition verification; and updating the system status based on the verification data, and confirming or canceling the early warning signal.

[0007] Secondly, an air quality monitoring device based on a medical clean space is provided, comprising: a first generation module, used to generate and output a corresponding early warning signal when a prediction model determines that an air quality anomaly will occur in a target area of ​​the medical clean space in the future; a first driving module, used to use the early warning signal as a task trigger command to drive a target mobile robot; a first receiving module, used to receive feedback data from the target mobile robot, wherein the feedback data is verification data generated by the target mobile robot in response to the task trigger command, based on the location information contained in the early warning signal, autonomously planning a path and moving to the target area, and performing environmental data collection and on-site condition verification; and a first updating module, used to update the system status based on the verification data, and confirm or cancel the early warning signal.

[0008] Thirdly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method.

[0009] Fourthly, a computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, implements the steps of the above-described method.

[0010] The aforementioned scheme for air quality monitoring in medical clean spaces, comprising a method, device, computer equipment, and storage medium, fundamentally improves upon traditional static monitoring methods by introducing a closed-loop intelligent monitoring mechanism of "prediction-drive-verification." First, by generating early warning signals before anomalies occur through a predictive model, the system's response node is moved from "post-event" to "pre-event," creating a valuable time window for proactive intervention and thus solving the problem of delayed risk detection. Second, the early warning signals are used as task instructions to directly drive mobile robots, achieving a leap from manual judgment to automatic dispatch, ensuring response speed and overcoming the inefficiency caused by reliance on manual labor. Subsequently, the target mobile robot autonomously plans its path and performs on-site verification, using its mobility to compensate for the blind spots of fixed sensors and accurately locating pollution sources, solving the problems of incomplete perception and inaccurate early warnings. Finally, the system status is updated based on the verification data transmitted back by the mobile robot, confirming or canceling the early warning signal, forming a self-verifying and self-optimizing intelligent closed loop. This effectively reduces the false alarm rate and ensures that subsequent decisions are based on real and accurate on-site information. In summary, through the synergistic effect of the aforementioned improvements, the air quality monitoring of medical clean spaces has been transformed from a traditional passive, static, and open-loop "alarm system" into an active, dynamic, and closed-loop "intelligent monitoring system." This enables accurate early warning, rapid response, and efficient closed-loop handling of air quality risks, thereby improving the efficiency of automatic air quality monitoring. Attached Figure Description

[0011] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 A schematic flowchart of an air quality monitoring method for medical clean spaces provided in an embodiment of the present invention; Figure 2 A schematic diagram of the first sub-process of the air quality monitoring method based on medical clean space provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the second sub-process of the air quality monitoring method based on medical clean space provided in an embodiment of the present invention; Figure 4 A schematic block diagram of an air quality monitoring device for medical clean spaces provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of a computer device according to an embodiment of the present invention; Figure 6This is another structural schematic diagram of a computer device according to one embodiment of the present invention. Detailed Implementation

[0013] 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, not all, of the embodiments of the present invention. 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.

[0014] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0015] This invention provides an air quality monitoring method based on medical clean spaces. The method can be applied to computer devices including but not limited to smartphones, tablets, desktop computers, servers, cloud computing, and mobile robots, and can be used for air quality monitoring in scenarios including but not limited to operating rooms, sterile preparation workshops, biological laboratories, and ICU wards.

[0016] The following detailed description of some embodiments of the present invention is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0017] Please see Figure 1 , Figure 1 This is a schematic flowchart illustrating an air quality monitoring method for medical clean spaces provided in an embodiment of the present invention. Figure 1 As shown, the method includes, but is not limited to, the following steps S11-S14: S11. When the prediction model determines that an abnormal air quality will occur in the target area of ​​the medical clean space in the future, a corresponding early warning signal is generated and output.

[0018] Explained, a predictive model refers to an algorithmic model trained on historical and real-time data that can infer future trends. In this embodiment of the invention, it refers to a time-series predictive model, which analyzes time-series data of air quality parameters (such as suspended particle concentration and microbial concentration) in medical clean spaces, learns their changing patterns, and thus predicts parameter values ​​within a specific future time window. The predictive model can be a statistical model (such as ARIMA), a machine learning model (such as an LSTM network), or a combination of both.

[0019] Medical clean spaces refer to enclosed spaces that effectively monitor and control airborne particulate matter, microorganisms, temperature, humidity, and pressure differences through technical engineering methods to meet the environmental requirements of specific medical activities (such as surgery, preparation, and intensive care). Examples include operating rooms, sterile preparation workshops, and ICU wards.

[0020] The target area refers to the specific spatial and geographical location within a medical clean space that is identified by the predictive model as potentially prone to air quality anomalies in the future. For example, it could be an area in an operating room where a surgical procedure is about to be performed, or the area around a patient bed in an ICU that is predicted to be at increased risk due to increased personnel movement.

[0021] Abnormal air quality indicates that key environmental parameters in the air deviate from their preset safety thresholds.

[0022] Based on the above concept and description, the system continuously collects first and second environmental data, performs spatiotemporal alignment and data fusion, forming an environmental data sequence reflecting the real-time status of the entire clean space. The first environmental data is collected by a network of fixed sensors deployed within the space, providing high-frequency baseline data covering key locations, such as particle concentration, temperature, humidity, and pressure difference. The second environmental data is collected by mobile robots performing routine inspections, providing high spatial resolution, supplementary dynamic data.

[0023] The fused environmental data sequence (including corresponding historical and real-time environmental data) is input into a pre-trained prediction model. The prediction model analyzes the inherent patterns and evolution trends of the data and outputs the predicted air quality parameters of each potential target area in the space within a future time window (e.g., the next 10-30 minutes). The predicted air quality parameters are compared with a preset safety threshold. When the prediction model predicts that the parameter value of a certain target area will continuously or repeatedly exceed the preset safety threshold in the future, it is determined that the area will experience air quality anomalies in the future.

[0024] Once an anomaly is determined to occur, the system immediately generates and outputs a structured warning signal. The warning signal is an instruction package containing key information, and its data structure includes at least: 1) Warning ID, representing a unique identifier; 2) Target area coordinates, representing the precise location in the digital twin model; 3) Predicted anomaly type, such as "exceeding the standard for suspended particle concentration"; 4) Predicted severity level, such as "medium"; 5) Expected occurrence time, such as "expected to occur in 15 minutes".

[0025] Thus, by introducing predictive analytics into medical clean environment monitoring, a fundamental shift has been achieved from "passive post-event alarms" to "proactive pre-event warnings."

[0026] S12. Use the warning signal as a task trigger command to drive the target mobile robot.

[0027] Explained, in this embodiment of the invention, a mobile robot refers to an intelligent device deployed within a medical cleanroom, possessing autonomous mobility, environmental perception, and data communication capabilities. It typically consists of a mobile chassis (such as an omnidirectional wheel chassis), a computing unit, a sensor suite (LiDAR, vision camera, high-precision particle counter, etc.), and a task execution mechanism (such as a microbial sampler). Its core function is to act as a "mobile perception node" and "on-site execution terminal" for the system, compensating for the blind spots of fixed sensor networks. The target mobile robot refers to a single mobile robot dynamically selected by the system from the mobile robot cluster during a specific early warning event, deemed most suitable for performing the "early warning verification" task.

[0028] Based on the above concept and setup, after receiving the warning signal generated in the aforementioned steps, the system (e.g., a digital twin platform) matches and generates a semantic "warning verification task instruction" from the preset task strategy library based on its contained information (such as target area and anomaly type). This instruction is a structured action guide, and its content goes far beyond a single coordinate point. For example, it includes "task type: warning verification; target location: operating table core area; execution mode: high-precision scanning mode; primary action: start particle counter and microbial sampler".

[0029] Then, based on the dynamic optimization strategy, the "target mobile robot" is determined from all online and available mobile robots. The dynamic optimization strategy is a multi-objective decision-making process, and the evaluation factors include, but are not limited to: 1) the proximity principle, calculating the estimated time for each robot to reach the target area; 2) capability matching, ensuring that the sensors carried by the robot can meet the task requirements (such as the need to equip a sampler for checking microbial early warning); 3) optimal state, prioritizing the selection of robots with sufficient power and low current task load.

[0030] Once the target mobile robot is identified, the system directly sends the encapsulated task instructions to the target mobile robot's control system via a wireless network to drive the target's movement. "Driven" here means triggering the start of the robot's autonomous task. After receiving the task instructions, the target mobile robot's control system will autonomously parse the instruction content and then trigger its task execution engine. This marks the robot's transition from a standby or interrupted state to being officially "driven" into the process of executing this early warning and verification task.

[0031] Thus, it realizes intelligent closed-loop linkage between the monitoring system and the execution equipment (i.e., mobile robots), transforming early warning information from "prompts that require human judgment" into "automated instructions that can directly drive actions in the physical world." This not only greatly shortens the cycle from risk discovery to on-site response, creating conditions for proactive intervention, but also eliminates the problem of response lag caused by delays or errors in human judgment, significantly improving the intelligence level and emergency response efficiency of the entire system.

[0032] S13. Receive feedback data from the target mobile robot. The feedback data is the verification data generated by the target mobile robot in response to the task triggering command, based on the location information contained in the warning signal, autonomously planning a path and moving to the target area, and performing environmental data collection and on-site condition verification.

[0033] Interpretatively, environmental data refers to the physical and chemical parameters collected by the target mobile robot within a target area using its onboard sensors to quantitatively assess air quality. These include, but are not limited to: 1) suspended particle concentration, such as the number of particles ≥0.5 μm and ≥5 μm, used to assess air cleanliness levels; 2) microbial aerosol concentration, such as the number of airborne colonies per unit volume of air, used to directly assess the risk of biocontamination; and 3) temperature, humidity, and pressure difference, as key environmental parameters affecting pollutant diffusion and cleanliness maintenance.

[0034] The on-site situation refers to the qualitative judgment made on the actual environmental safety status of the target area by real-time analysis of the collected environmental data and combined with the robot's visual observation. Its core is to verify the authenticity of the warning and locate the source of risk. For example, "confirmed abnormality" can mean that a specific pollution source has been found and located, such as a large number of particles being continuously released from the equipment's heat dissipation vent. "confirmed normal" can mean that after a comprehensive scan, all environmental parameters in the area are within the safe threshold, and the warning is determined to be a false alarm.

[0035] "Continued risk" indicates that no fixed source of pollution has been found, but periodic exceedances of local concentrations have been detected due to gatherings of people.

[0036] As described above, the target mobile robot uses its current position as the starting point and the target area coordinates in the warning signal as the ending point. Based on a static environmental map (including walls, fixed equipment, etc.) stored in systems such as digital twin models, it calculates a collision-free optimal global path using search algorithms such as A or D Lite. During movement along the global path, the robot continuously uses its LiDAR and depth vision sensors to perceive the environment ahead in real time, dynamically constructing a local obstacle map. When dynamic obstacles (such as medical personnel or mobile medical vehicles) are detected, its control system immediately initiates a local path replanning algorithm (such as Dynamic Window Method (DWA)) to generate detour instructions, ensuring a safe and smooth arrival at the target area. Upon arrival, the mobile robot autonomously switches to a "high-precision monitoring mode," maximizing the sampling frequency of its onboard high-precision laser particle counter and microbial aerosol sampler to capture instantaneous, subtle changes in pollution. Furthermore, the robot does not perform static measurements but executes a pre-defined coverage scanning path (such as a serpentine path) within the target area to construct a high-resolution spatial distribution map of pollutants in that area.

[0037] During the scanning process, the spatial gradient of pollutant concentration is analyzed in real time. If an abnormally high local concentration is found, the "concentration gradient climbing tracking algorithm" is automatically activated. The algorithm adjusts its movement path along the direction of gradually increasing concentration until it locates the highest concentration point, i.e., the potential pollution source. At the same time, the visual sensor is used to assist in observation to confirm the type of pollution source (such as equipment overheating, door not sealing properly, etc.), thus completing the final verification of the on-site situation.

[0038] This endows the system with precise on-site insight and source tracing capabilities. Through the autonomous movement and intelligent scanning of robots, the single "point warning" is expanded into a spatial and refined diagnosis of the entire "area." This not only verifies the authenticity of the warning but, more importantly, accurately locates the root cause of pollution, providing a decisive basis for subsequent precise handling (such as notifying personnel to clean specific equipment). This achieves a leap from "knowing there is a problem" to "knowing where the problem is and what it is," greatly improving the pertinence and effectiveness of control measures.

[0039] S14. Based on the verification data, update the system status and confirm or cancel the warning signal.

[0040] Explanatory verification data refers to the structured data packets returned to the system by the target mobile robot after completing the on-site verification task. These data packets are not only the raw environmental readings, but also the "on-site diagnostic conclusions" after preliminary processing by the robot. The core contents include: 1) Environmental data, such as high-precision, high-spatial-resolution data on particle and microbial concentrations collected in the target area; 2) Pollution source location data, such as the specific coordinates and types of pollution sources (e.g., equipment heat dissipation, personnel activities), and related image / video evidence; 3) Robot trajectory data, the complete movement path of the robot during the scanning and source tracing process, used to reproduce its actions in the digital twin model; 4) On-site condition judgment, qualitative conclusions drawn based on the above data, such as "confirmed abnormality - pollution source located", "confirmed abnormality - cause unknown" or "confirmed normality - warning lifted".

[0041] Based on the above concept and description, the digital twin platform and other systems receive and parse the verification data transmitted back by the target mobile robot. Then, the verification data is fused with other information in the system (such as real-time data from fixed sensors and original information from early warning signals) to form a complete and three-dimensional data view of the early warning event. Based on the fused information, using environmental and trajectory data from the verification data, the dynamic air quality map in the digital twin model is updated. For example, the location of the pollution source discovered by the robot is marked as a highlighted red area, and the surrounding pollutant concentration cloud map is updated. Furthermore, if the on-site condition in the verification data is determined to be "confirmed,"... If an "abnormality" is detected, the system will officially update the warning signal status from "Warning in Progress" to "Confirmed Warning." This operation will trigger a higher-level alarm (such as escalating the audible alarm or sending an emergency notification to the person in charge), and may also include a push notification with handling suggestions provided by the verification data (such as "Please clean the cooling fan of the anesthesia machine located at coordinates [X,Y]"). If the verification data indicates that everything in the target area is normal (i.e., the on-site situation is determined to be "Confirmed Normal"), the system will deactivate the warning signal. This operation will restore the system status to "Normal Monitoring" and record this event as a "false alarm." This record can be used to optimize the prediction model and reduce the future false alarm rate.

[0042] Therefore, by feeding back the robot's on-site inspection results to the system core, the entire system can become more and more accurate and intelligent as it runs, ultimately achieving a leap from single-time automated response to long-term autonomous evolution.

[0043] This invention fundamentally improves traditional static monitoring methods by introducing a closed-loop intelligent monitoring mechanism of "prediction-drive-verification". First, a predictive model generates early warning signals before anomalies occur, shifting the system's response from "after-the-fact" to "before-the-fact," creating a valuable time window for proactive intervention and solving the problem of delayed risk detection. Second, the early warning signals serve as direct task commands to drive mobile robots, achieving a leap from manual judgment to automatic dispatch, ensuring response speed and overcoming the inefficiency caused by reliance on manual intervention. Subsequently, the target mobile robot autonomously plans its path and performs on-site verification, using its mobility to compensate for the blind spots of fixed sensors and accurately locating pollution sources, solving the problems of incomplete perception and inaccurate early warnings. Finally, the system status is updated based on the verification data transmitted back by the mobile robot, confirming or canceling the early warning signal, forming a self-verifying and self-optimizing intelligent closed loop. This effectively reduces the false alarm rate and ensures that subsequent decisions are based on real and accurate on-site information. In summary, through the synergistic effect of the aforementioned improvements, the air quality monitoring of medical clean spaces has been transformed from a traditional passive, static, and open-loop "alarm system" into an active, dynamic, and closed-loop "intelligent monitoring system." This enables accurate early warning, rapid response, and efficient closed-loop handling of air quality risks, thereby improving the efficiency of automatic air quality monitoring.

[0044] In one embodiment, please refer to Figure 2 , Figure 2 This is a schematic diagram of the first sub-process of the air quality monitoring method based on medical clean spaces provided in an embodiment of the present invention. Figure 2 As shown, the predictive model identifies areas within the medical cleanroom that are likely to experience air quality anomalies in the future, including: S21. Obtain first environmental data, second environmental data, and historical environmental data sequence of the target area, wherein the first environmental data is collected by a fixed sensor network, and the second environmental data is collected by a mobile robot. S22. The first environmental data, the second environmental data and the historical environmental data sequence are spatiotemporally aligned and fused to form a fused environmental data sequence; S23. Input the fused environmental data sequence into the pre-trained time series prediction model and output the predicted air quality parameters of the target area in the future time window. S24. When the predicted value of the air quality parameter continuously exceeds the preset safety threshold and reaches the preset number of times, it is determined that the target area will experience air quality anomalies in the future.

[0045] Interpretationally, a historical environmental data sequence represents a set of historical records of first and second environmental data related to a target area stored in a database. This sequence is used to enable predictive models to learn the periodic and trend-based changes in environmental parameters under the influence of specific events (such as staff shift changes or equipment startup).

[0046] Pre-trained time series prediction models refer to algorithmic models that are pre-trained based on a large amount of historical data and are capable of processing time series data and predicting its future values. For example, long short-term memory network models can capture the precursor patterns that lead to air quality deterioration from complex environmental data.

[0047] Air quality parameter prediction values ​​represent the specific numerical predictions of core environmental parameters (such as 0.5μm particle concentration and airborne bacteria concentration) of the target area within a specific future time window (such as the next 15-30 minutes) output by the prediction model. These values ​​are the direct basis for determining abnormal risks.

[0048] Based on the above concept and setup, this embodiment is the core of achieving forward-looking early warning. Through multi-source data fusion and intelligent analysis, it elevates monitoring from "perceiving the present" to "predicting the future." The main steps are as follows: The system collects first environmental data from fixed sensors and second environmental data from a mobile robot in parallel. For effective fusion, spatiotemporal alignment is required. Spatial alignment means unifying the acquisition locations of all data to the global coordinate system of the digital twin model. Temporal alignment means unifying the timestamps of all data to a standard time axis and interpolating and resampling non-uniformly spaced data (such as data collected during robot movement) to form a data sequence with a uniform time step. After spatiotemporal alignment, the current data is concatenated with the historical environmental data sequence to form a complete fused environmental data sequence, which combines the continuity of fixed points with the spatial details of moving points.

[0049] The aforementioned fused environmental data sequence is input into a pre-trained time-series prediction model. This model analyzes the inherent patterns in the sequence and outputs predicted air quality parameters for the target area within a future time window. The system compares this predicted value sequence with a preset safety threshold. Furthermore, to avoid misjudgments caused by single-point fluctuations, stricter judgment conditions are set: only when the predicted value continuously exceeds the threshold and the number of times it exceeds the threshold reaches a preset number (for example, if five consecutive data points exceed the threshold within the predicted 20 minutes), is it finally determined that "air quality anomalies will occur in the future." This ensures the high reliability of the early warning.

[0050] This invention achieves early and accurate insight into air quality risks in medical clean spaces by constructing an intelligent prediction closed loop based on multi-source fusion data. First, by spatiotemporally aligning and fusing fixed first environmental data with moving second environmental data, a fused data sequence with both temporal continuity and spatial detail is formed, providing a comprehensive and accurate data foundation for prediction. Second, deep analysis of the fused data sequence using a pre-trained time-series prediction model captures precursory patterns leading to anomalies from complex environmental changes, achieving a leap from "monitoring what has happened" to "predicting what will happen," providing a core technical means to solve the problem of delayed risk detection. Finally, an intelligent criterion of "continuously exceeding and reaching a preset number of times" is used for anomaly determination, effectively filtering out instantaneous and isolated interference signals, significantly reducing the system's false alarm rate, ensuring the accuracy and reliability of early warning information, and enabling subsequent emergency responses to be based on highly reliable early warnings. In summary, through the synergistic effect of the above data fusion, intelligent prediction, and accurate determination, crucial decision-making basis is provided for the proactive and precise control of the entire system.

[0051] In one embodiment, please refer to Figure 3 , Figure 3 This is a schematic diagram of the second sub-process of the air quality monitoring method based on medical clean spaces provided in an embodiment of the present invention. Figure 3 As shown, using the warning signal as a task trigger command to drive the target mobile robot includes: S31. The warning signal is sent to the digital twin model corresponding to the medical clean space; S32. The digital twin model generates semantic verification task instructions based on the target area location indicated by the warning signal, the anomaly type, and the preset task strategy library; S33. The digital twin model dynamically selects the optimal mobile robot from the mobile robot cluster as the target mobile robot based on the current location, power status and current task load of each mobile robot. S34. Assign the verification task instruction to the target mobile robot.

[0052] Interpretatively, a digital twin model refers to a virtual entity created through digital modeling techniques that maintains real-time synchronization and interaction with a medical clean space in the physical world in terms of geometry, physics, rules, and behavior. It is not only a three-dimensional visualization model, but also a system simulation and decision-making center that integrates and merges real-time data from fixed sensor networks, the status and position of mobile robots, and environmental equipment information.

[0053] The target area location refers to the location of the target area, and also represents the spatial coordinate information contained in the warning signal that may cause anomalies in the future. This location corresponds to a specific three-dimensional spatial range in the digital twin model, and is the destination of the mission execution.

[0054] The anomaly type indicates the category of potential risk identified by the warning signal, such as "excessive concentration of suspended particles," "risk of microbial contamination," or "abnormal pressure difference." Different types of anomalies correspond to different verification focuses and resource requirements.

[0055] A pre-defined task strategy library represents a set of "if-then" rules or strategy mapping tables stored in the digital twin model. It defines the logic of what specific verification tasks should be generated for different anomaly types and target area characteristics. For example, the strategy library can define that if the anomaly type is "microbial contamination risk", then the task instruction must include "start microbial sampler".

[0056] Semantic verification task instructions represent high-level task descriptions generated by digital twin models that can be understood and executed by mobile robots. Unlike simple coordinate points, they are structured instructions that encapsulate task intent, action objectives, and constraints. For example, a semantic instruction might be: {Task Type: Early Warning Verification, Target Location: [Coordinates XYZ], Core Action: High-Precision Particle Scan & Microbial Sampling, Execution Mode: Coverage Scan}.

[0057] Based on the above concept and setup, after receiving an early warning signal, the digital twin model immediately analyzes it, extracting key attributes such as the target area location and anomaly type. Subsequently, the model queries a pre-set task strategy library, matching the attributes with the rules in the library to generate semantic verification task instructions. This process transforms the abstract early warning into a work order containing specific operational requirements that can be directly executed by the robot.

[0058] The digital twin model, while generating task instructions, performs a global state assessment of the mobile robot swarm to dynamically select the optimal mobile robot as the target robot. This selection is a multi-objective optimization decision-making process, with evaluation factors including: 1) proximity principle: calculating the estimated arrival time of each robot to the target area and prioritizing the robot with the shortest time; 2) capability matching: ensuring that the sensors on the robot can meet the task requirements (e.g., robots need to be equipped with samplers to verify microbial early warning); 3) optimal state: prioritizing robots with sufficient battery power and low current task load (e.g., in standby or performing interruptible tasks). The system uses a preset weighted scoring algorithm to comprehensively score the candidate robots, ultimately selecting the robot with the highest comprehensive score as the target mobile robot.

[0059] Finally, the digital twin model accurately sends the generated semantic verification task instructions to the control system of the selected target mobile robot through a wireless communication network. After receiving the instructions, the target mobile robot's autonomous task engine is triggered, marking that it has been officially "driven" and has begun to execute this verification task.

[0060] This invention, through intelligent scheduling using a digital twin model, automates and optimizes the early warning response process. First, by converting early warning signals into semantic verification task instructions, the complex on-site decision-making process is pre-emptively implemented and embedded in the system. This ensures that the task instructions themselves contain execution strategies, overcoming the inefficiency and inaccuracy of traditional methods that rely on manual issuance of specific instructions, thus guaranteeing the standardization and normalization of response actions. Second, by dynamically selecting the optimal target mobile robot based on multi-dimensional status information (location, power level, load), precise scheduling and efficient utilization of system resources (robots) are achieved. This ensures response speed while avoiding task allocation conflicts or task interruptions due to poor robot status, thereby improving the overall system throughput and reliability. In summary, the entire system can automatically, quickly, and accurately allocate early warning tasks to the most suitable execution unit, providing crucial support for the core invention's "prediction-drive-verification" closed loop, and greatly enhancing the system's collaborative intelligence and overall response efficiency.

[0061] In one embodiment, the target mobile robot, in response to the task triggering command, autonomously plans a path and moves to the target area based on the location information contained in the warning signal, including: After receiving the verification task instruction, the target mobile robot autonomously calls and switches to the early warning verification mode from its task mode library; Based on the global environment map provided by the digital twin model and the location of the target area, an initial global path is planned; During the movement, the target mobile robot uses its lidar and depth vision sensor to perceive the surrounding environment in real time and construct a local obstacle map. When a dynamic obstacle is detected, the local path is replanned in real time based on the local obstacle map to avoid the obstacle and continue moving towards the target area.

[0062] Explained, the task mode library represents a set of control strategies and behavioral logics pre-installed within the target mobile robot's control system. Each mode defines specific behavioral parameters for the robot for different task types, such as movement speed, sensor sampling frequency, and navigation aggression. For example, the library may include "routine inspection mode," "emergency tracing mode," and the "early warning verification mode" involved in the embodiments of this invention.

[0063] The early warning verification mode is a specific mode in the task mode library designed specifically for performing early warning response tasks. When this mode is invoked, the robot automatically configures its system to prioritize task execution efficiency and data accuracy. For example, its behavioral parameters may be set to: medium movement speed (to balance efficiency and stability), the highest sensor sampling frequency (to capture subtle changes in contamination), and allow for brief pauses to avoid obstacles (to ensure personnel safety).

[0064] A global environment map is a two-dimensional or three-dimensional map containing static structural information of a medical cleanroom, provided by a digital twin model and stored within the robot. This map includes the precise locations of invariant elements such as walls, fixed equipment, and doorways, providing the robot with a macroscopic, prior environmental framework for global path calculation.

[0065] A local obstacle map is a dynamically constructed map that reflects only temporary, movable obstacles as the target mobile robot scans its surrounding environment (e.g., within a 180-degree radius in front, or within 10 meters) in real time during its movement, using its onboard LiDAR and depth vision sensors. This map complements the global environment map and is used to address dynamically changing environments.

[0066] Depth vision sensors are visual sensors (such as RGB-D cameras and stereo vision cameras) capable of acquiring distance information of objects within their field of view. They can not only identify objects like ordinary cameras, but also calculate precise 3D distance data between objects and robots. This data can then be fused with LiDAR data to accurately construct local obstacle maps, especially useful in identifying transparent or reflective surfaces (such as glass doors) to compensate for the limitations of LiDAR.

[0067] Based on the above concept and setup, after receiving the verification task instruction from the digital twin model, the target mobile robot first autonomously calls and switches to the early warning verification mode from its task mode library. This switching action reconfigures the robot's control system parameters (such as PID controller parameters and sensor data acquisition frequency) to a state that matches this mode, enabling it to enter the optimal working state for performing the early warning verification task. Based on the switched mode, the robot's path planner, using its current position as the starting point and the target area location contained in the early warning signal as the ending point, combines the global environment map provided by the digital twin model and uses search algorithms such as A (A-Star) or D (D-Star) to calculate an optimal initial global path from the starting point to the ending point, avoiding all static obstacles. This path serves as the macroscopic guiding route for the robot's movement. The robot begins to move along the global path. During the movement, it continuously uses LiDAR for two-dimensional or three-dimensional scanning, supplemented by a depth vision sensor for object recognition and distance perception. By fusing and matching scan data from different times and locations, a local obstacle map centered on itself is constructed and updated in real time. When a dynamic obstacle is detected (such as a medical worker or mobile medical vehicle suddenly appearing in a corridor), the robot's obstacle avoidance module immediately marks the obstacle on a local obstacle map. Subsequently, based on this real-time map, the system uses local planning algorithms such as Dynamic Window (DWA) or Temporal Flexible Band (TEB) to perform real-time local path replanning within milliseconds, generating a smooth local path segment that safely avoids the current obstacle. Once the obstacle is bypassed, the robot quickly returns to the initial global path and continues moving towards the target area.

[0068] This invention, through its embodiment, endows the robot with a high degree of environmental awareness and autonomous decision-making capabilities, ensuring the reliable execution of early warning response tasks. First, by setting up a task mode library and an early warning verification mode, the robot can adaptively adjust its behavioral strategies according to different tasks, ensuring that its performance configuration is optimal when performing critical early warning tasks, thus improving the professionalism of task execution and the quality of data collection. Second, by employing a hierarchical path planning strategy combining "global planning and local replanning," it utilizes prior environmental information provided by the digital twin model to ensure the optimality of the macroscopic path, while effectively addressing the inherent dynamism and uncertainty of the medical environment through real-time perception and local decision-making. This fundamentally solves the safety and maneuverability challenges of robot movement in complex crowd environments. In summary, the target mobile robot is no longer simply an automated device following a fixed route, but an autonomous intelligent entity capable of intelligently coping with complex real-world environments and safely and efficiently completing point-to-point movement. This provides crucial mobility assurance for the core invention to achieve "precise driving" and "on-site verification," significantly improving the practicality and robustness of the entire system in real medical scenarios.

[0069] In one embodiment, performing environmental data collection and on-site condition verification includes: Once the target mobile robot arrives at the target area, it activates its onboard high-precision particle counter and microbial aerosol sampler to collect environmental data of the target area at a sampling frequency higher than that of the conventional inspection mode. Simultaneously, the target mobile robot is controlled to perform a comprehensive path scan within the target area; During the scanning process, the spatial distribution of pollutant concentration is monitored in real time. If an abnormally high local concentration is detected, a tracking algorithm based on concentration gradient climbing is activated to autonomously adjust the movement path to locate potential pollution sources.

[0070] Explained, a high-precision particle counter is a precision instrument that uses the principle of laser scattering to measure and count the number of suspended particles of different sizes (such as ≥0.5μm and ≥5μm particles) in a unit volume of air in real time and at high resolution.

[0071] A microbial aerosol sampler is a device that can actively extract a quantitative amount of air and impact the microbial particles (such as bacteria and fungi) onto a specific culture medium for subsequent culturing and counting.

[0072] A higher sampling frequency than the regular inspection mode indicates that, in the early warning and verification mode, the sensor data acquisition interval is significantly shortened to capture transient and subtle changes in pollution. For example, in the regular inspection mode, the particle counter may sample once every 10 seconds; while in this mode, it is set to sample once per second to obtain a data stream with higher temporal resolution.

[0073] Coverage path scanning means that the target mobile robot moves within the target area according to a preset, systematic motion trajectory (such as a serpentine or spiral path) to ensure that its sensors perform comprehensive and uniform spatial sampling of the area, thereby constructing a cloud map of pollutant concentration distribution in the area.

[0074] The concentration gradient-based tracking algorithm is an intelligent search algorithm that simulates chemical directional behavior. Its core principle is to calculate the spatial gradient of concentration (i.e., the direction of the fastest concentration increase) by continuously comparing the pollutant concentration values ​​of the current measurement point with those of several previous measurement points. Based on this, the robot's movement direction is autonomously adjusted so that it moves along the direction of the concentration gradient until it locates the point where the concentration can no longer increase, i.e., the pollution source.

[0075] Based on the above concept and setup, when the target mobile robot arrives at the target area, it first activates its onboard high-precision particle counter and microbial aerosol sampler, setting the sampling frequency to a higher frequency than the conventional inspection mode. Simultaneously, the robot controls itself to perform a comprehensive path scan within the target area. This dual action ensures high-resolution capture of environmental data in both spatial and temporal dimensions. During the scan, the robot's data processing unit monitors and records its geographical location and corresponding pollutant concentration readings in real time. The system analyzes these spatiotemporal data points to identify local concentration anomalies. Specifically, the instantaneous concentration value at the current point is compared with the current average concentration and historical background concentration of the entire target area. When the concentration value at that point is consistently and significantly higher than the surrounding environment (e.g., exceeding the average by more than two standard deviations), it is identified as a local concentration anomaly. Once a local concentration anomaly is identified, the robot immediately initiates a concentration gradient-based tracking algorithm. This algorithm controls the robot to pause its preset comprehensive path and, based on the real-time collected concentration data, autonomously calculate and move a short distance in the direction of the largest concentration gradient, then sample, calculate, and move again. Through this iterative process of "perception-decision-movement", the system eventually tracks down to the location with the highest concentration that no longer changes significantly, thereby accurately locating potential pollution sources.

[0076] This invention, through its embodiment, endows the robot with sophisticated on-site detection and intelligent diagnostic capabilities, enabling substantial verification and root cause localization of early warning signals. First, by activating high-precision sensors and performing high-frequency, comprehensive scanning, the robot obtains high-resolution data far exceeding that of fixed-point sensors, revealing the spatial heterogeneity of pollutants and providing a solid data foundation for accurately verifying on-site conditions. Second, by introducing a tracking algorithm based on concentration gradient ascent, the robot possesses the intelligent behavior of actively exploring and locating pollution sources, elevating the system's capability from "confirming a problem" to "locating the root cause of the problem," thus addressing the pain points of low efficiency and reliance on experience in manual investigation. In summary, this makes the target mobile robot a powerful terminal integrating a "high-precision mobile monitoring station" and an "intelligent on-site investigator," providing indispensable and decisive on-site insights for the core invention to achieve a closed loop from risk warning to precise handling, greatly enhancing the value and effectiveness of the entire system's response.

[0077] In one embodiment, updating the system status based on the verification data and confirming or deactivating the warning signal includes: The digital twin model receives verification data transmitted back by the target mobile robot, the verification data including environmental data, pollution source location data, and robot trajectory data; The verification data is used to update the dynamic air quality map in the digital twin model; If the verification data confirms the existence of a pollution source or abnormal condition, the system status will be updated to "Confirmation Warning"; If the verification data indicates that the air quality in the target area is normal, the warning signal is lifted, and the system status is updated to "risk cleared".

[0078] Interpretatively, environmental data refers to a set of pre-processed air quality parameters collected by the target mobile robot within the target area. These parameters mainly include high-precision particle concentration readings and microbial sampling status, and are the direct quantitative basis for condition assessment.

[0079] Pollution source location data refers to the spatial coordinate information of the pollution source determined by the target mobile robot through intelligent source tracing algorithm, and may include image or video data for supporting evidence. The data accurately points out the root location of the risk, such as {pollution source coordinates: [X,Y,Z], source type: equipment heat dissipation, evidence: image link}.

[0080] Robot trajectory data represents the spatial coordinate sequence of the complete path traversed by the target mobile robot during its coverage scanning and pollution source tracing operations. This data is used to recreate the robot's movement route in a digital twin model and spatially correlate it with the collected environmental data.

[0081] A dynamic air quality map is a graphical representation of the real-time spatial distribution of pollutant concentrations within a medical cleanroom, rendered in the form of color clouds or isosurfaces within a digital twin model. It is a dynamically updated, global situational awareness view.

[0082] Based on the above concept and description, the digital twin model receives the structured verification data packet returned by the target mobile robot, parses it, and extracts the environmental data, pollution source location data, and robot trajectory data.

[0083] Using the analyzed verification data, the dynamic air quality map in the digital twin model is updated: the path covered by the robot trajectory data is highlighted in the model; environmental data collected along the trajectory (such as particle concentration values) is calculated using a spatial interpolation algorithm (such as Kriging interpolation) to generate a continuous pollutant concentration distribution surface covering the target area; this newly generated distribution surface is then merged and overlaid with the existing data on the map, thereby refreshing the dynamic air quality map to reflect the latest environmental conditions verified by the robot. If pollution source location data exists, a prominent pollution source identifier is generated at the corresponding coordinates.

[0084] Based on the updated map and verification data, if the verification data confirms the existence of a pollution source or abnormal situation (e.g., persistently exceeding particle concentration limits and the specific source has been located), the system will officially update the warning signal's status from "Under Verification" to "Confirmed Warning." This status will trigger a higher-level alarm notification and may automatically generate a work order containing the location and type of pollution source. If the verification data indicates that the air quality in the target area is normal (i.e., all environmental parameters are within safe thresholds and no pollution source has been found), the system will deactivate the warning signal and update the system status to "Risk Deactivated." Simultaneously, this event can be recorded as a "false alarm" case for subsequent optimization of the prediction model's accuracy.

[0085] This invention, through the construction of a decision-making closed loop based on real feedback, achieves precise management and self-optimization of the system state. First, by updating the dynamic air quality map using environmental data, pollution source location data, and robot trajectory data transmitted by the robot, the system's global situational awareness remains highly consistent with the actual on-site conditions. This overcomes the situational distortion problem caused by traditional systems relying solely on sparse fixed-point data, providing management personnel with a basis for decision-making. Second, the automated decision-making process for "confirming" or "cancelling" warning signals based on verification data ensures accurate judgment for each warning. This avoids the disruption to operations caused by continuous false alarms and ensures that real risks can be immediately escalated and addressed, fundamentally solving the shortcomings of traditional open-loop systems where "no one takes action after an alarm" or "responds slowly." In summary, the entire system possesses intelligent characteristics of self-verification, self-decision-making, and self-evolution. It not only completes the closed loop of a single warning response but also lays the foundation for long-term accuracy and reliability improvement through accumulated verification data, ultimately achieving a transformation from an automated tool to an intelligent collaborative partner.

[0086] In one embodiment, the method further includes: When an abnormal air quality is detected by the fixed sensor network, the digital twin model generates semantic emergency tracing task instructions. Based on the current state of each mobile robot, the emergency tracing task instruction is dynamically assigned to the corresponding target mobile robot; After receiving the emergency source tracing task instruction, the target mobile robot autonomously switches to the emergency source tracing mode and moves to the abnormal area to perform rapid location and source tracing operations for the pollution source.

[0087] Interpretatively, an emergency source tracing task instruction refers to a semantic task instruction with the highest response priority generated by a digital twin model upon receiving an alarm for current air quality anomalies. The core objective of this instruction is to "quickly locate and confirm the pollution source." Its content structure is similar to that of an early warning verification instruction, but it typically includes an emergency identifier and may omit certain fine-grained scanning parameters to prioritize response speed. For example: {Task Type: Emergency Source Tracing, Target Location: [Coordinates of the Anomaly Area], Priority: Emergency, Core Action: Rapid Source Tracing}.

[0088] Emergency tracing mode is a special operating mode pre-installed in the mobile robot's task mode library, designed specifically for responding to sudden current anomalies. When switching to this mode, the robot will adjust its behavioral parameters to maximize response speed, such as: using the highest safe movement speed, simplifying unnecessary path smoothing, and prioritizing the invocation of computationally more efficient fast tracing algorithms.

[0089] An abnormal area refers to the specific spatial range where current air quality parameters (such as particle concentration) have exceeded a preset safety threshold, as directly monitored by a fixed sensor network. This area is typically a preliminary investigation range centered on the location of the fixed sensor that triggered the alarm, combined with spatial topology.

[0090] Based on the above concept and setup, when the real-time data stream from the fixed sensor network shows that the data at a certain location continuously exceeds the safety threshold, the system determines that the current air quality is abnormal and locks down the abnormal area. The digital twin model then intervenes, generating semantic emergency source tracing task instructions from a pre-set strategy library based on the anomaly type and location. Next, the model dynamically allocates the emergency instruction to the most suitable target mobile robot using an optimization strategy similar to that used in early warning responses, based on the current status of each mobile robot (such as location, battery level, and whether the task can be interrupted). Upon receiving the instruction, the target mobile robot immediately and autonomously switches to emergency source tracing mode from its task mode library, entering an "emergency deployment" state. Subsequently, based on the abnormal area location information in the instruction, the robot initiates path planning and moves to the area in the most efficient way to perform rapid pollution source location and tracing operations.

[0091] This invention, through the construction of an independent and efficient emergency response channel, enhances the system's ability to handle risks across the entire spectrum. Firstly, by establishing dedicated emergency tracing task instructions and emergency tracing modes, the system establishes a standardized rapid response process for occurring emergencies, ensuring that resources are prioritized for addressing confirmed risks and overcoming the task congestion and response delays that can occur with single-process systems. Secondly, it complements and synergizes with the aforementioned early warning and verification process, creating a system that simultaneously addresses potential risks and existing events, significantly enhancing its robustness and comprehensiveness in complex and dynamic medical environments. In summary, this makes the entire monitoring system a complete solution that is both forward-looking and highly responsive, ensuring that regardless of the form (potential or sudden) of the risk, the system provides a matching, efficient, and automated response, thereby comprehensively protecting the air safety of medical clean spaces.

[0092] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0093] In one embodiment, an air quality monitoring device based on a medical cleanroom is provided, which corresponds one-to-one with the air quality monitoring method based on a medical cleanroom described in the above embodiments. Please refer to [link / reference]. Figure 4 , Figure 4 This is a schematic block diagram of an air quality monitoring device for medical clean spaces provided as an embodiment of the present invention. Figure 4 As shown, the air quality monitoring device 40 based on medical clean spaces includes a first generation module 41, a first driving module 42, a first receiving module 43, and a first updating module 44. The detailed descriptions of each functional module are as follows: The first generation module 41 is used to generate and output a corresponding early warning signal when the prediction model determines that an air quality abnormality will occur in the target area of ​​the medical clean space in the future. The first drive module 42 is used to use the warning signal as a task triggering command to drive the target mobile robot. The first receiving module 43 is used to receive feedback data from the target mobile robot. The feedback data is the verification data generated by the target mobile robot in response to the task triggering command, based on the location information contained in the warning signal, autonomously planning a path and moving to the target area, and performing environmental data collection and on-site condition verification. The first update module 44 is used to update the system status based on the verification data and to confirm or cancel the warning signal.

[0094] In one embodiment, the first generation module 41 includes: A first acquisition submodule is used to acquire first environmental data, second environmental data, and a historical environmental data sequence of the target area, wherein the first environmental data is collected by a fixed sensor network, and the second environmental data is collected by a mobile robot; a first fusion submodule is used to perform spatiotemporal alignment and fusion of the first environmental data, the second environmental data, and the historical environmental data sequence to form a fused environmental data sequence; a first output submodule is used to input the fused environmental data sequence into a pre-trained time series prediction model and output the predicted air quality parameters of the target area within a future time window; a first determination submodule is used to determine that the target area will experience air quality anomalies in the future when the predicted air quality parameters continuously exceed a preset safety threshold and reach a preset number of times.

[0095] In one embodiment, the first driving module 42 includes: The first sending submodule is used to send the warning signal to the digital twin model corresponding to the medical clean space; the digital twin model, based on the target area location indicated by the warning signal, the anomaly type, and the first generation submodule, is used to preset a task strategy library and generate semantic verification task instructions; the first selection submodule is used by the digital twin model to dynamically select the optimal mobile robot from the mobile robot cluster as the target mobile robot based on the current location, power status, and current task load of each mobile robot; the first allocation submodule is used to allocate the verification task instructions to the target mobile robot.

[0096] In one embodiment, the first receiving module 43 includes: The first invocation submodule is used for the target mobile robot to autonomously call and switch to the early warning verification mode from its task mode library after receiving the verification task instruction; the first planning submodule is used to plan an initial global path based on the global environment map provided by the digital twin model and the location of the target area; the first construction submodule is used to construct a local obstacle map by using the lidar and depth vision sensor on the target mobile robot to perceive the surrounding environment in real time during the movement; the second planning submodule is used to replan the local path in real time based on the local obstacle map when dynamic obstacles are detected, so as to avoid obstacles and continue to move towards the target area.

[0097] In one embodiment, the first receiving module 43 includes: The first acquisition submodule is used to activate the high-precision particle counter and microbial aerosol sampler carried by the target mobile robot after it arrives at the target area, and collect environmental data of the target area at a sampling frequency higher than that of the conventional inspection mode. The first control submodule is used to simultaneously control the target mobile robot to perform a comprehensive path scan within the target area. The first adjustment submodule is used to monitor the spatial distribution of pollutant concentration in real time during the scanning process. If an abnormally high local concentration is found, a tracking algorithm based on concentration gradient climbing is activated to autonomously adjust the movement path to locate potential pollution sources.

[0098] In one embodiment, the first update module 44 includes: The first receiving submodule is used for the digital twin model to receive verification data transmitted back by the target mobile robot, the verification data including environmental data, pollution source location data, and robot trajectory data; the first updating submodule is used to update the dynamic air quality map in the digital twin model using the verification data; the second updating submodule is used to update the system status to "confirmed warning" if the verification data confirms the existence of a pollution source or abnormal situation; the third updating submodule is used to cancel the warning signal and update the system status to "risk cleared" if the verification data indicates that the air quality in the target area is normal.

[0099] In one embodiment, the air quality monitoring device 40 further includes: The second generation module is used to generate semantic emergency tracing task instructions from the digital twin model when the current air quality is detected by the fixed sensor network; the first allocation module is used to dynamically allocate the emergency tracing task instructions to the corresponding target mobile robots according to the current state of each mobile robot; the first movement module is used to enable the target mobile robot to autonomously switch to emergency tracing mode after receiving the emergency tracing task instructions and move to the abnormal area to perform rapid location and tracing of pollution sources.

[0100] This invention provides an air quality monitoring device for medical clean spaces, fundamentally improving upon traditional static monitoring methods by introducing a closed-loop intelligent monitoring mechanism of "prediction-drive-verification". First, a predictive model generates early warning signals before anomalies occur, shifting the system's response from "post-event" to "pre-event," creating a valuable time window for proactive intervention and solving the problem of delayed risk detection. Second, the early warning signals serve as task instructions to directly drive a mobile robot, achieving a leap from manual judgment to automatic dispatch, ensuring response speed and overcoming the inefficiency caused by reliance on manual intervention. Subsequently, the target mobile robot autonomously plans its path and performs on-site verification, using its mobility to compensate for the blind spots of fixed sensors and accurately locating pollution sources, solving the problems of incomplete perception and inaccurate early warnings. Finally, the system status is updated based on the verification data transmitted back by the mobile robot, confirming or canceling the early warning signal, forming a self-verifying and self-optimizing intelligent closed loop. This effectively reduces the false alarm rate and ensures that subsequent decisions are based on real and accurate on-site information. In summary, through the synergistic effect of the aforementioned improvements, the air quality monitoring of medical clean spaces has been transformed from a traditional passive, static, and open-loop "alarm system" into an active, dynamic, and closed-loop "intelligent monitoring system." This enables accurate early warning, rapid response, and efficient closed-loop handling of air quality risks, thereby improving the efficiency of automatic air quality monitoring.

[0101] Specific limitations regarding air quality monitoring devices for medical clean spaces can be found in the above-mentioned limitations on air quality monitoring methods for medical clean spaces, and will not be repeated here. Each module in the aforementioned air quality monitoring device for medical clean spaces can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0102] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 5 As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile and / or volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with external clients via a network connection. When the computer program is executed by the processor, it implements the functions or steps of a server-side method for air quality monitoring in a medical cleanroom.

[0103] In one embodiment, a computer device is provided, which may be a client, and its internal structure diagram may be as follows: Figure 6 As shown, the computer device includes a processor, memory, network interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with an external server via a network connection. When the computer program is executed by the processor, it implements the functions or steps of a client-side method for air quality monitoring in a medical cleanroom.

[0104] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the air quality monitoring method based on medical clean spaces described in the above embodiments.

[0105] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the air quality monitoring method for medical clean spaces described in the above embodiments.

[0106] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0107] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0108] The software tools or components not belonging to our company that appear in the embodiments of this invention are merely illustrative examples and do not represent actual use.

[0109] The data collection in this embodiment of the invention complies with the requirements of relevant laws and regulations, such as China's Personal Information Protection Law, GDPR (General Data Protection Regulation of the European Union), or information security standards of other countries and regions.

[0110] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for monitoring air quality based on a medical clean space, characterized in that, The method comprises the following steps: When it is judged by the prediction model that air quality anomaly will occur in the target area of the medical clean space in the future, a corresponding early warning signal is generated and output; The early warning signal is used as a task trigger instruction to drive the target mobile robot; Receiving return data from the target mobile robot, the return data being verification data generated after the target mobile robot autonomously plans a path and moves to the target area based on the position information contained in the early warning signal in response to the task trigger instruction, collects environment data, and verifies the on-site situation; Based on the verification data, update the system state and confirm or cancel the early warning signal.

2. The medical clean space based air quality monitoring method as claimed in claim 1, wherein, Judging by the prediction model that air quality anomaly will occur in the target area of the medical clean space in the future, comprising: Obtaining first environment data, second environment data and historical environment data sequence of the target area, wherein the first environment data is collected by a fixed sensor network, and the second environment data is collected by a mobile robot; The first environment data, the second environment data and the historical environment data sequence are spatio-temporally aligned and fused to form a fused environment data sequence; The fused environment data sequence is input into a pre-trained time series prediction model to output air quality parameter prediction values of the target area in a future time window; When the air quality parameter prediction values continuously exceed the preset safety threshold and reach a preset number of times, it is determined that air quality anomaly will occur in the target area in the future.

3. The medical clean space based air quality monitoring method as claimed in claim 1, wherein, The early warning signal is used as a task trigger instruction to drive the target mobile robot, comprising: The early warning signal is sent to the digital twin model corresponding to the medical clean space; The digital twin model generates a semantic verification task instruction based on the target area position, anomaly type and preset task strategy library indicated by the early warning signal; The digital twin model dynamically selects the optimal mobile robot as the target mobile robot from the mobile robot cluster according to the current position, power state and current task load of each mobile robot; The verification task instruction is distributed to the target mobile robot.

4. The medical clean space based air quality monitoring method as claimed in claim 3, wherein, The target mobile robot autonomously plans a path and moves to the target area based on the position information contained in the early warning signal in response to the task trigger instruction, comprising: After receiving the verification task instruction, the target mobile robot autonomously calls and switches to the early warning verification mode from its task mode library; Based on the global environment map provided by the digital twin model and the target area position, an initial global path is planned; During the movement, the laser radar and depth vision sensor carried by the target mobile robot are used to perceive the surrounding environment in real time, and a local obstacle map is constructed; When a dynamic obstacle is detected, real-time local path re-planning is performed based on the local obstacle map to avoid obstacles and continue moving to the target area.

5. The medical clean space based air quality monitoring method of claim 3 or 4, wherein, Collecting environment data and verifying on-site situation, comprising: When the target mobile robot reaches the target area, start the high-precision particle counter and microbial aerosol sampler carried by the target mobile robot to collect environmental data of the target area at a higher sampling frequency than the regular inspection mode; At the same time, control the target mobile robot to perform a covering path scan in the target area; During the scanning process, the spatial distribution of pollutant concentration is monitored in real time, and if a locally abnormally high point is found, a tracking algorithm based on concentration gradient climbing is started to adjust the moving path autonomously to locate the potential pollution source.

6. The medical clean space-based air quality monitoring method of claim 3 or 4, wherein, Based on the verification data, update the system state and confirm or cancel the early warning signal, including: The digital twin model receives the verification data returned by the target mobile robot, including environmental data, pollution source positioning data, and robot trajectory data; The dynamic air quality map in the digital twin model is updated using the verification data; If the verification data confirms the existence of a pollution source or abnormal condition, the system state is updated to "confirmed early warning"; If the verification data indicates that the air quality of the target area is normal, the early warning signal is canceled, and the system state is updated to "risk removed".

7. The medical clean space based air quality monitoring method of claim 3 or 4, wherein, The method further includes: When the current air quality is monitored to be abnormal through the fixed sensor network, the digital twin model generates a semanticized emergency traceability task instruction; According to the current state of each mobile robot, the emergency traceability task instruction is dynamically assigned to the corresponding target mobile robot; After receiving the emergency traceability task instruction, the target mobile robot autonomously switches to the emergency traceability mode and moves to the abnormal area to perform rapid positioning and traceability operation of the pollution source.

8. An air quality monitoring device for a medical clean space, comprising: Including: The first generation module is configured to generate and output a corresponding early warning signal when it is determined through the prediction model that the target area of the medical clean space will have air quality abnormalities in the future; The first driving module is configured to drive the target mobile robot using the early warning signal as a task triggering instruction; The first receiving module is configured to receive return data from the target mobile robot, which is verification data generated by the target mobile robot after autonomously planning a path and moving to the target area based on the location information contained in the early warning signal in response to the task triggering instruction, performing collection of environmental data and verification of on-site conditions; The first update module is configured to update the system state based on the verification data and confirm or cancel the early warning signal.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the steps of the air quality monitoring method based on a medical clean space according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, the computer program comprising instructions that, when executed by a computer, cause the computer to perform the method of any one of claims 1 to 9. The computer program is executed by the processor to implement the steps of the air quality monitoring method based on a medical clean space according to any one of claims 1 to 7.

Images