A big data management and intelligent evaluation system for hospital environment air quality

Abstract

The present application relates to the field of intelligent regulation and energy saving technology of hospital environment air quality, in particular to a big data management and intelligent evaluation system for hospital environment air quality; comprising multi-modal data acquisition, data processing, dynamic transmission modeling, dynamic Bayesian network risk modeling and forward-looking risk hedging and energy optimization modules; the system collects environment and people flow data, calculates dynamic air transmission coefficient, and constructs a dynamic Bayesian network model to calculate the cross-zone transmission risk probability; the core is that when the risk probability exceeds the threshold value, the system will actively solve the multi-objective optimization problem with the goal of minimizing energy consumption to generate optimal HVAC control instructions; the present application realizes the change from lagging evaluation to forward-looking risk hedging, and can actively identify and regulate the risk before the pollution exceeds the standard.

Description

Technical Field

[0001] This invention relates to the field of intelligent control and energy-saving technology for hospital ambient air quality, specifically a big data management and intelligent assessment system for hospital ambient air quality. Background Technology

[0002] In current hospital environmental management, HVAC systems are the core facilities for ensuring air quality. However, existing HVAC systems generally suffer from static, passive, and high energy consumption problems. They typically adopt fixed operating strategies or only activate purification after detecting excessive pollutant concentrations, which is a delayed and passive response. To ensure the absolute safety of critical areas such as the ICU, the system often adopts a one-size-fits-all strategy of high-power operation, resulting in high overall energy consumption. At the same time, traditional static models cannot cope with dynamic disturbances in the hospital environment, especially the instantaneous impact of personnel movement and the opening and closing of doors and windows on air transmission, causing model prediction failures.

[0003] Therefore, how to integrate multimodal dynamic data, accurately assess and proactively predict the risk of cross-regional spread of pollutants caused by dynamic disturbances, and on this basis achieve synergistic optimization between ensuring the safety of key areas and energy conservation and consumption reduction of the system, is a technical problem that urgently needs to be solved. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a big data management and intelligent assessment system for hospital ambient air quality. Specifically, the technical solution of this invention includes:

[0005] The multimodal data acquisition module is used to collect environmental status data and dynamic pedestrian flow data in real time.

[0006] The data processing module is used to normalize the real-time pressure difference data in the dynamic pedestrian flow data and the environmental status data to obtain normalized pedestrian activity indicators and normalized real-time pressure difference.

[0007] The dynamic transmission modeling module is used to combine the preset baseline aerodynamic coupling coefficient, the normalized pedestrian activity index, the normalized real-time pressure difference, and the gate state data in the dynamic pedestrian data, and calculate the dynamic air transmission coefficient through a preset dynamic disturbance function.

[0008] The dynamic Bayesian network risk modeling module is used to calculate the probability of cross-regional transmission risk based on the dynamic air transport coefficient and the pollutant concentration data in the environmental state data.

[0009] The forward-looking risk hedging and energy consumption optimization module is used to: compare the cross-regional propagation risk probability with a preset risk tolerance threshold; generate a normal operation signal in response to the cross-regional propagation risk probability being less than the risk tolerance threshold; and solve a multi-objective optimization problem with the objective of minimizing a preset energy consumption cost function and the constraint that the simulated risk probability is lower than a preset constraint value in response to the cross-regional propagation risk probability being greater than or equal to the risk tolerance threshold, thereby calculating the optimal HVAC control command.

[0010] Preferably, the environmental status data includes: particulate matter concentration, microbial concentration, carbon dioxide concentration, temperature and humidity, and differential pressure data between key areas.

[0011] Preferably, the dynamic pedestrian flow data includes: pedestrian density, flow direction, average dwell time, and door status data.

[0012] Preferably, the preset baseline aerodynamic coupling coefficient is calibrated by an aerodynamic map construction module; wherein, the aerodynamic map construction module defines the hospital functional areas as nodes, defines the physical connections between areas as edges, and calibrates the baseline aerodynamic coupling coefficient through computational fluid dynamics simulation or tracer gas experiments.

[0013] Preferably, the preset constraint value is the product of the preset risk tolerance threshold and the preset safety redundancy coefficient.

[0014] Preferably, the simulated risk probability is calculated through "What-if" deduction;

[0015] The "What-if" deduction includes:

[0016] Based on the HVAC control command vector to be evaluated and the current differential pressure data, the new differential pressure after regulation is predicted by the HVAC-pressure response model;

[0017] Substituting the new pressure difference into the preset dynamic disturbance function, the simulated dynamic transmission coefficient is calculated;

[0018] The simulated dynamic transmission coefficients are used, and Bayesian network forward inference is re-executed based on the current state to calculate the simulated risk probability.

[0019] Preferably, the probability of cross-regional transmission risk is defined as: the conditional probability that, given the observed occurrence of a specific pollutant event in the source region and the current observational evidence, the pollutant concentration in the target critical area will exceed a preset cleanliness threshold within a future time step.

[0020] Preferably, the preset risk tolerance threshold is set differently based on the medical functional importance and cleanliness level requirements of the target area.

[0021] Preferred options also include:

[0022] The HVAC control execution interface module is used to translate the normal operating signals or the optimal HVAC control commands into physical control signals of the underlying HVAC actuators and send them down for execution.

[0023] Compared with the prior art, the present invention has the following beneficial effects:

[0024] 1. This invention achieves forward-looking probabilistic prediction of the risk of cross-regional transmission of pollutants by constructing a dynamic Bayesian network risk model and combining it with real-time collected multimodal data; the system can proactively identify risks before pollution actually exceeds the standard, rather than passively responding after the standard is exceeded, realizing the transformation from lagging assessment to forward-looking risk hedging;

[0025] 2. This invention solves the technical deficiency of traditional static models in failing to reflect the dynamic disturbances of hospitals by introducing a dynamic transmission modeling module; the system combines a pre-calibrated baseline aerodynamic coupling coefficient with a real-time dynamic disturbance function, which integrates people flow, real-time pressure difference between key areas and door opening and closing status, thereby accurately and dynamically quantifying the instantaneous air transmission capacity between areas;

[0026] 3. This invention resolves the core conflict between ensuring the safety of critical areas and saving energy in the system in a hospital environment. When the predicted risk exceeds the threshold, the system does not adopt a one-size-fits-all high-power strategy. Instead, it solves a multi-objective optimization problem with the goal of minimizing energy consumption cost and the condition that the simulated risk probability is lower than the safety constraint. Under the premise of ensuring safety, it achieves coordinated energy-saving optimization of the HVAC system.

[0027] 4. This invention achieves refined management by setting differentiated risk tolerance thresholds, allowing for targeted measures based on specific areas. The system allows different risk tolerance levels to be set according to the medical function importance and cleanliness level requirements of different areas, thereby precisely allocating the control resources of the HVAC system to the truly necessary areas and avoiding the waste of high energy consumption of the entire system to protect a few key areas. Attached Figure Description

[0028] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0029] Figure 1 This is a structural diagram of the system of the present invention. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0031] Example 1:

[0032] Please see Figure 1 A big data management and intelligent assessment system for hospital ambient air quality, comprising:

[0033] The multimodal data acquisition module is used to collect environmental status data and dynamic pedestrian flow data in real time.

[0034] The data processing module is used to normalize the real-time pressure difference data in dynamic pedestrian flow data and environmental status data to obtain normalized pedestrian flow activity indicators and normalized real-time pressure difference.

[0035] The dynamic transmission modeling module is used to combine the preset baseline aerodynamic coupling coefficient, normalized pedestrian activity index, normalized real-time pressure difference and gate state data in dynamic pedestrian data, and calculate the dynamic air transmission coefficient through the preset dynamic disturbance function.

[0036] The dynamic Bayesian network risk modeling module is used to calculate the probability of cross-regional transmission risk based on the dynamic air transport coefficient and pollutant concentration data in the environmental status data.

[0037] The forward-looking risk hedging and energy consumption optimization module is used to: compare the probability of cross-regional propagation risk with a preset risk tolerance threshold; generate a normal operation signal in response to the probability of cross-regional propagation risk being less than the risk tolerance threshold; and solve a multi-objective optimization problem with the objective of minimizing a preset energy consumption cost function and the constraint that the simulated risk probability is lower than a preset constraint value in response to the probability of cross-regional propagation risk being greater than or equal to the risk tolerance threshold, and calculate the optimal HVAC control command.

[0038] This invention provides a big data management and intelligent assessment system for hospital ambient air quality. The system aims to address the static, passive, and high-energy-consumption problems of existing hospital HVAC systems. By integrating multimodal data and aerodynamic modeling, it achieves forward-looking prediction of the risk of cross-regional pollutant transmission and coordinated energy consumption optimization. In a preferred embodiment, the system includes: a multimodal data acquisition module, a data processing module, a dynamic transmission modeling module, a dynamic Bayesian network risk modeling module, and a forward-looking risk hedging and energy consumption optimization module.

[0039] The multimodal data acquisition module aims to acquire multi-source heterogeneous data characterizing the dynamic changes of the hospital's internal environment in real time and comprehensively. This data forms the foundation for subsequent dynamic modeling and risk prediction. In this embodiment, the module is implemented through a sensor network deployed in various functional areas of the hospital. The data collected by this module is mainly divided into two categories: environmental status data and dynamic pedestrian flow data. Environmental status data refers to parameters characterizing the physical environment and air quality. In this embodiment, it specifically includes: particulate matter concentration in each area. Such as PM2.5, PM10, microbial concentration, Concentration, temperature, humidity, and differential pressure data between key areas that have a critical impact on airflow. Among them, differential pressure data This refers to the real-time pressure difference between region i and region j, which characterizes the driving force of airflow. It is generated through real-time data collected by differential pressure sensors deployed on both sides of doors or partitions. Dynamic pedestrian flow data refers to data caused by human activity that strongly disturbs airflow. In this embodiment, it specifically includes: pedestrian density in each region obtained through anonymous video analysis or Wi-Fi detection. Airflow direction, average residence time, and gate state data that have a decisive impact on inter-regional air exchange. Among them, gate state data This refers to the door between regions i and j being open at time t. Or close The source of this is the status sensor installed on the door frame;

[0040] The data processing module aims to clean, align, and standardize the raw data acquired by the multimodal data acquisition module, providing dimensionally consistent and meaningful input for subsequent dynamic models. In this embodiment, this module focuses on dynamic pedestrian flow data, such as pedestrian density. Real-time differential pressure data in environmental status data, such as Perform a normalization operation; its output is a normalized pedestrian activity index. This refers to the density of people. Relative to a design baseline value Such as the dimensionless ratio of the maximum design capacity of the area; its normalized real-time differential pressure output. This refers to real-time differential pressure. Relative to a standard differential pressure reference Such as the dimensionless ratio of standard differential pressure values ​​required by design specifications;

[0041] The dynamic air transport modeling module aims to quantify the air exchange capacity between different functional areas, such as from a corridor to the ICU, in real time and dynamically. This is one of the key innovations that distinguishes this invention from traditional static models. The core task of this module is to calculate the dynamic air transport coefficient. To characterize this dynamic exchange capability, this embodiment introduces a dynamic air transport coefficient. It is a semi-empirical model based on physical intuition, and its calculation method is as follows:

[0042] ;

[0043] in, :exist At any moment, from the source region To the target area The dynamic air transport coefficient; its technical meaning is characterized by... Within the time step, from District transferred to The air volume in the area accounts for A relevant transfer ratio of the region volume; a dimensionless parameter; calculated in real time by this module;

[0044] : The preset baseline aerodynamic coupling coefficient; its physical meaning is to represent the area under baseline conditions such as nighttime unmanned operation and standard HVAC operation. air on the region The influence weight; dimensionless; this coefficient is pre-calibrated by the aerodynamic map construction module during the system initialization phase through CFD simulation or tracer gas experiments.

[0045] : A preset dynamic perturbation function; its technical motivation lies in the static Unable to reflect real-time disturbances such as pedestrian flow and pressure fluctuations, this invention uses this function to dynamically modulate the baseline coefficient; both the function and its output are dimensionless modulation factors, for example, a value of 1.2 indicates a 20% increase in transmission capacity; its specific form can be, for example, a rule-based fuzzy logic system, or an exponential function and its internal parameters such as... This is the core parameter to be adjusted in this invention; as a non-limiting specific example, the preset dynamic perturbation function The following exponential function form can be used:

[0046] ;

[0047] in, When the function value is 1, the function value is greater than When it is 0; These are the parameters to be calibrated; this form specifically characterizes how the flow of people in the source and target regions, the normalized pressure difference between the two regions, and the door opening status all modulate and enhance the baseline aerodynamic coupling coefficient.

[0048] Its calibration source is: during the system calibration phase, using historical datasets containing historical pollutant concentrations. Historical flow of people Historical pressure difference and historical status Through machine learning methods, for example, function with As undetermined parameters, they are substituted into the mass balance model described below to construct a complete state predictor. This is achieved by minimizing the concentration predicted by the model. Compared with historical concentrations observed The error between them, such as the root mean square error, is used to solve in reverse and optimize to obtain the parameters. ;

[0049] Normalized pedestrian activity indicators; calculated in real time by the data processing module;

[0050] Normalized real-time differential pressure; calculated in real time by the data processing module;

[0051] Door status data in dynamic pedestrian flow data; dimensionless, 0 for closed, 1 for open; acquired in real time by the multimodal data acquisition module; using this formula, the dynamic transmission modeling module successfully transmits static, pre-calibrated building maps. With real-time disturbance sources of pedestrian flow Pressure difference Gate status The dynamic combination enables accurate characterization of instantaneous air transport capability, overcoming the deficiency of static models in the background technology that cannot reflect dynamic disturbances;

[0052] The Dynamic Bayesian Network Risk Modeling Module (DBN) aims to leverage the capabilities provided by the Dynamic Transport Modeling Module. Construct a probabilistic reasoning model to predict the near future. The module addresses the risk of pollutants spreading across regions and causing pollution levels to exceed standards in target areas; it constructs a DBN where the state variables for each time slice include pollutant concentration data for each region. The core of this module is to utilize To define the state transition model of DBN, i.e. Time's up Pollutant concentration at time The probability distribution; the discrete time step The selection of [the appropriate method / mechanism] is one of the key steps in realizing this invention, and its selection requires balancing the timeliness of the physical response with the stability of the calculation. It should be less than the average duration of typical disturbances in the environment, but long enough to smooth sensor noise and allow the DBN to complete one forward inference; in this embodiment, The preferred value range is 30 seconds to 5 minutes; in this embodiment, this transition relationship is based on a discrete time step. To characterize the simplified mass balance model on Expected value:

[0053] ;

[0054] in, Target area exist Expected pollutant concentration at any given time;

[0055] : The concentration residual coefficient of the region, characterizing exist The residual proportion of HVAC supply air, return air, and natural decay over a period of time; dimensionless; its source is: Compared with the number of air exchanges in zone j and time step Related, for example, can be approximated as ;

[0056] Cross-regional propagation term; this is the key to the invention, it utilizes dynamic air transport coefficient. Concentration in the source region , characterizing in Within a time period, from District Intake The increase in concentration contributed by pollutants in the area;

[0057] Internal pollution source contribution item, by Pollution sources within the area, such as those emitted by people's breathing and equipment, are present. The concentration contribution generated within the time step; its source is: a baseline source intensity pre-set based on regional functions such as the number of patients in the ward and medical activities. or through the area concentration The rate of change is estimated in real-time through inversion; as a non-limiting specific example, when using... When the rate of change of concentration is inverted, this Yuanqiang It can be based on the following Estimation of the inversion formula for mass balance:

[0058] ;

[0059] in, It is a region volume, It is a region Total ventilation volume It is air supply Concentration; after estimating Afterwards, if If this refers to the concentration contribution of particulate matter or microorganisms, then we can further assume its physical source strength. and Yuanqiang They are linearly correlated, that is ,in, Let be the release rate coefficient calibrated using historical data; then It can be calculated as The DBN module utilizes this transfer model, combined with real-time evidence input from the multimodal data acquisition module. like , , Using standard DBN inference algorithms such as particle filtering or forward propagation, the probability of cross-regional propagation risk is iteratively calculated and finally output. To achieve probabilistic reasoning, the DBN module assumes that, given... At any given time, Concentration at time It follows a Gaussian distribution; its mean is That is Its variance This characterizes the uncertainty of the model; It can be pre-calibrated, or It can also be a function that dynamically changes with the intensity of the real-time disturbance; based on this probability distribution The DBN module can then calculate That is, the probability of cross-regional transmission risk as defined in this invention. ;

[0060] The Forward-Looking Risk Hedging and Energy Optimization module is the system's decision-making and control center. Its purpose is to achieve intelligent regulation from passive response to proactive prevention, based on the risks predicted by the DBN module, while ensuring the safety of all areas of the hospital, especially critical areas such as the ICU, and minimizing energy consumption. The module's operational logic is divided into two branches:

[0061] Risk Comparison with Routine Operation: Real-time Module Monitoring and compare it with the preset risk tolerance threshold. Compare; This refers to the hospital management based on the region. The importance of medical functions, such as the ICU, has a very low threshold, while the corridor has a higher threshold, which is a pre-set tolerable risk limit; in response to In other words, if the system predicts that the future risks are within an acceptable safety range, the system determines that no intervention is needed and generates a normal operating signal, such as maintaining the current HVAC operating status or implementing a normal energy-saving strategy.

[0062] Risk hedging and optimization decision-making: responsive to That is, the system predicts the future If a pollution exceeding the standard event is highly likely to occur within a certain timeframe, a proactive hedging mechanism will be immediately triggered. In this case, the module will not adopt a static strategy based on maximum power, but will instead solve a multi-objective optimization problem to calculate the optimal HVAC control command. This optimization problem is another core aspect of the present invention, and in this embodiment it is expressed as a constrained optimization problem:

[0063]

[0064] st Preset constraint values

[0065] in, : The HVAC control command vector to be optimized; these are the variables to be optimized, for example... That is, for the source region and target area The control parameters of the relevant HVAC equipment;

[0066] : Preset energy consumption cost objective function; this function estimates the energy consumption based on HVAC equipment models such as fan energy consumption curves. The additional energy consumption resulting from the instruction is measured in kWh; its model parameters are derived from the equipment manual or actual calibration.

[0067] The core of simulating risk probability constraints; this is a What-if simulation probability, representing the probability under the assumption that... After adjustment, the system re-predicts future risks;

[0068] Preset constraints: the minimum safety threshold that the optimization must meet; the module iteratively solves the problem using well-known optimization algorithms such as gradient descent and particle swarm optimization until a solution is found that satisfies the constraints. <Preset constraint value and Minimal and optimal HVAC control commands ;

[0069] This embodiment, through the collaborative work of the aforementioned multimodal data acquisition module, data processing module, dynamic transmission modeling module, dynamic Bayesian network risk modeling module, and forward-looking risk hedging and energy consumption optimization module, constructs a complete technical closed loop from real-time perception, dynamic modeling, probabilistic prediction to optimized control; unlike the passive response of traditional HVAC systems, which only initiate purification after pollution exceeds the standard, this system... and The model can identify risks before pollutants actually spread across regions, realizing a shift from passive assessment to proactive risk hedging. More importantly, this system resolves the core conflict between safety and energy conservation in hospital environments. It abandons the one-size-fits-all strategy of keeping the entire system running at maximum power for extended periods to ensure ICU safety, and instead achieves precise hedging through proactive risk hedging and constrained optimization of energy consumption modules: that is, only hedging when necessary. Take necessary measures This ensures that the cleanliness of critical areas such as the ICU and operating rooms meets standards such as ISO14644, while greatly reducing the overall energy consumption of the HVAC system, thus achieving the goals of refined, dynamic, and energy-saving hospital environmental management.

[0070] Example 2:

[0071] Environmental status data include: particulate matter concentration, microbial concentration, carbon dioxide concentration, temperature and humidity, and differential pressure data between key areas.

[0072] Dynamic pedestrian flow data includes: pedestrian density, flow direction, average dwell time, and door status data.

[0073] This embodiment further specifies the data collected by the multimodal data acquisition module in Embodiment 1. Based on Embodiment 1, environmental state data is one of the key inputs of the multimodal data acquisition module, and its purpose is to comprehensively characterize the physical environment and air quality inside the hospital. In this embodiment, it specifically includes: particulate matter concentrations such as PM2.5 and PM10 and microbial concentrations such as those obtained by sedimentation bacteria method or airborne bacteria sampler for assessing air cleanliness.

[0074] Carbon dioxide concentration used to assess fresh air volume and indoor air freshness Temperature and humidity for assessing thermal comfort; differential pressure data between key areas serving as core inputs to the dynamic transport modeling module and data processing module; simultaneously, dynamic crowd flow data provides the main real-time sources of aerodynamic disturbances; in this embodiment, it specifically includes: crowd density obtained through anonymous video analysis or Wi-Fi detection. Flow direction, average residence time; and gate state data that have a decisive influence on the aerodynamic coupling relationship between regions. ;

[0075] This embodiment utilizes environmental condition data and dynamic pedestrian flow data, especially pressure difference. Gate state Through fusion acquisition, this invention provides a basis for calculating the dynamic air transport coefficient in the subsequent dynamic transmission modeling module. All necessary real-time physical inputs; this ensures The model can realistically and dynamically reflect the complex hospital microenvironment jointly determined by personnel activities, door opening and closing disturbances, and HVAC system operation, which is essential for accurate risk prediction. The physical basis of it.

[0076] Example 3:

[0077] The preset baseline aerodynamic coupling coefficient is calibrated through the aerodynamic map construction module; wherein, the aerodynamic map construction module defines the functional areas of the hospital as nodes and the physical connections between areas as edges, and calibrates the baseline aerodynamic coupling coefficient through computational fluid dynamics simulation or tracer gas experiments.

[0078] This embodiment focuses on the key input parameter of the dynamic transmission modeling module in Embodiment 1—the preset baseline aerodynamic coupling coefficient. —Specific explanation of the source and calibration method; based on Example 1, the preset baseline aerodynamic coupling coefficient The calibration was performed during the initialization phase of the system deployment using a proprietary aerodynamic mapping module. The purpose of this module is to characterize the inherent, static aerodynamic coupling relationships of the hospital building structure, providing a benchmark for subsequent dynamic corrections.

[0079] The aerodynamic mapping module is based on the hospital's architectural blueprints and HVAC design drawings. It defines each functional area of ​​the hospital, such as the ICU, wards, and corridors, as a node in the mapping. All possible physical connections between these areas, such as doors, windows, ventilation ducts, and even building gaps, are defined as edges in the mapping. This module uses computational fluid dynamics (CFD) simulation or tracer gas experiments to analyze each edge in the mapping. From the region arrive Calibration is performed; for example, in a CFD simulation, the simulation assumes baseline conditions such as no one being present at night and standard HVAC operation, and that a unit concentration of tracer gas is released in region i within a specified simulation time step. The concentration in region j was then measured to calibrate the baseline aerodynamic coupling coefficient. ; It is a dimensionless parameter, and its physical meaning is: under baseline conditions, the region air on the region The influence weight or transmission ratio;

[0080] This embodiment introduces an aerodynamic map construction module and pre-calibrates it. This invention has significant advantages in technical implementation; it successfully separates complex, computationally intensive CFD simulations or expensive and time-consuming tracer gas experiments from real-time control systems that require millisecond-level responses. As a static benchmark, it greatly reduces the real-time computational complexity of the dynamic transmission modeling module; this module only needs to... Based on this, perform lightweight multiplication operations. You can get This ensures high accuracy while achieving dynamic modeling with high timeliness, meeting the real-time requirements of forward-looking hedging.

[0081] Example 4:

[0082] The preset constraint value is the product of the preset risk tolerance threshold and the preset safety redundancy coefficient.

[0083] This embodiment is a specific definition of the preset constraint values ​​in the optimization problem solved by the forward-looking risk hedging and energy consumption optimization module in Embodiment 1; in the optimization problem of Embodiment 1, the constraint condition is st <Preset constraint value; This constraint value is the safety baseline that the system must reach after implementing hedging control; To ensure safety after control and to avoid frequent oscillations of the system at the risk threshold, this embodiment specifically defines the preset constraint value; The preset constraint value is set as a preset risk tolerance threshold.> With a preset safety redundancy factor The product of; where, This refers to a dimensionless, adjustable parameter less than 1, set by system administrators based on actual operational results such as hedging success rate; for example, it can be set to 0.8. Its function is to provide a safety buffer, ensuring that the optimizer finds... The corresponding simulated risk probability It must be significantly lower than, for example, a 20% lower risk tolerance threshold. ;

[0084] This embodiment introduces a safety redundancy factor. By defining preset constraint values, the hedging decisions of this system are more robust and stable; it avoids the HVAC system being affected by small risk fluctuations. Just in The frequent start-stop or adjustment caused by the nearby environment results in oscillations; this hysteresis characteristic in regulation ensures a higher safety margin and indirectly reduces the additional energy consumption and HVAC equipment wear caused by frequent adjustments.

[0085] Example 5:

[0086] The simulated risk probability is calculated using What-if deduction;

[0087] What-if deduction includes:

[0088] Based on the HVAC control command vector to be evaluated and the current differential pressure data, the new differential pressure after regulation is predicted by the HVAC-pressure response model;

[0089] Substitute the new pressure difference into the preset dynamic disturbance function to calculate the simulated dynamic transmission coefficient;

[0090] The simulated risk probability is calculated by using the simulated dynamic transmission coefficients and re-performing Bayesian network forward inference based on the current state.

[0091] This embodiment demonstrates how the forward-looking risk hedging and energy consumption optimization module in Embodiment 1 calculates the core parameters of its constraints to simulate risk probabilities when solving optimization problems. Detailed explanation; simulated risk probability It is solved through a What-if deduction process; the purpose of the What-if deduction is to quickly predict, through model simulation, what would happen if a certain HVAC control command vector were executed, without actually executing HVAC control. Future risks What will it be? In this embodiment, the What-if deduction includes the following sequential steps: an optimizer such as a particle swarm optimization algorithm generates an HVAC control command vector to be evaluated. For example, {ICU supply fan speed +10%, corridor exhaust fan frequency +2%}; the system calls an HVAC-pressure response model. For example, a reduced-order model based on a fluid resistance network or simplified by CFD, based on and current Pressure difference data at time Predict the new differential pressure that will be generated after this instruction is executed. As a non-restrictive concrete example, this HVAC-pressure response model It could be a simplified linear model based on a fluid resistance network; for example, assuming With the region air volume and region exhaust volume Related; HVAC control commands This will cause changes in air volume The model can be represented as:

[0092]

[0093] in, and It is the regional pressure response coefficient calibrated through historical data regression or fluid simulation;

[0094]

[0095] The system will use this new pressure difference. Normalization Substitute into the preset dynamic perturbation function In the middle, and assuming the flow of people Gate state exist The value remains constant over time, thus allowing for the calculation of a simulated dynamic transmission coefficient. ;

[0096]

[0097] The dynamic Bayesian network risk modeling module uses this simulated dynamic transmission coefficient. Replace the real And based on the current state, such as The Bayesian network forward inference algorithm is re-executed to finally calculate the simulated risk probability. ;

[0098] This embodiment utilizes the What-if reasoning process to achieve the physical control of HVAC at the algorithm level. With aerodynamic models and probabilistic risk model They are tightly coupled; this allows the optimizer of the forward-looking risk hedging and energy optimization modules to evaluate hundreds of different control strategies in milliseconds, far faster than the actual physical response. Safety effect and energy consumption This ensures that the optimal HVAC control commands output truly achieve the dual goals of safety and energy efficiency, which is the key technology for achieving forward-looking optimization.

[0099] Example 6:

[0100] The probability of cross-regional transmission risk is defined as the conditional probability that, given the observed occurrence of a specific pollutant event in the source region and the current observational evidence, the pollutant concentration in the target critical area will exceed a preset cleanliness threshold within a future time step.

[0101] This embodiment provides a detailed explanation of the technical meaning and probability definition of the core output of the dynamic Bayesian network risk modeling module in Embodiment 1—the cross-regional propagation risk probability; based on Embodiment 1, the cross-regional propagation risk probability... It has a clear, conditional probability-based definition; its purpose is not to measure the current concentration, but to predict the probability of future exceedances; in this embodiment, the probability of cross-regional transmission risk is strictly defined as:

[0102]

[0103] in, Conditional probability; This refers to risk events monitored by the system, specifically targeting critical areas. Such as the concentration of pollutants in the ICU Future time step The cleanliness level exceeded its preset cleanliness threshold. ; : Preset cleanliness threshold; its function is to determine the area The legal or regulatory basis for determining whether pollution exceeds the standard; its source is the upper limit of cleanliness set for the ISO 7 level of the area, such as the ICU, according to national regulations or international standards such as ISO 14644.

[0104] The first condition refers to... At that moment, the source region was observed. For example, a specific contaminant event occurred in the corridor; for instance, the multimodal data acquisition module detected the source area. A sudden surge of people or a momentary high concentration detected by its sensors ;

[0105] The second condition refers to... All current observational evidence observed at any given time; this includes real-time pedestrian flow data, real-time differential pressure data, and door status data for all areas provided by the multimodal data acquisition module;

[0106] This embodiment, through such a precise definition of the conditional probability of cross-regional transmission risk, achieves a shift from passive concentration assessment to proactive risk prediction; the system no longer focuses on the current concentration. Whether the limit has already been exceeded is too late now; the question is now in the current disturbance. Next, future How likely is it that the standard will be exceeded? This is a probability-based, forward-looking risk indicator. This makes the decision-making basis of the forward-looking risk hedging and energy consumption optimization module more scientific and robust, and can effectively cope with various uncertain disturbances in the hospital environment, such as the flow of people and the opening and closing of doors.

[0107] Example 7:

[0108] The preset risk tolerance threshold is set differently based on the medical function importance and cleanliness level requirements of the target area.

[0109] This embodiment focuses on the key parameter used by the forward-looking risk hedging and energy consumption optimization module in Embodiment 1 to trigger hedging decisions—the preset risk tolerance threshold. —Detailed explanation of the basis for setting; Based on Example 1, the preset risk tolerance threshold It is not a uniform, one-size-fits-all value for the entire hospital, but rather a differentiated value set according to the heterogeneous needs of different functional areas of the hospital; This refers to the system's target area. The maximum tolerable level, i.e., the maximum level that does not trigger intervention. Probability value; in this embodiment, this threshold is determined by hospital management or infection control experts based on the target area. The importance of medical functions and cleanliness requirements are predetermined; for example, for ICUs, operating rooms, or negative pressure isolation wards, their medical functions are crucial, and the cleanliness requirements are extremely high, such as ISO 5 or negative pressure requirements, therefore their... It will be set to an extremely low level, such as 0.01%; this means that as long as the system predicts a one in ten thousand chance of exceeding the limit in the ICU. All of these will immediately trigger the highest priority hedging mechanism; however, for public corridors, their functional importance is relatively low, and their cleanliness requirements are also lower. It can then be set relatively high, for example, 5%;

[0110] This embodiment perfectly meets the practical needs of hospitals with complex functional zoning and varying cleanliness standards; the system no longer requires excessive control of the corridor HVAC system to ensure ICU safety, which is precisely the reason for the huge energy waste in the background technology; through differentiation... The forward-looking risk hedging and energy consumption optimization module enables refined management tailored to specific regions, thus conserving precious energy resources. Prioritizing the protection of truly critical areas, it achieves synergistic optimization of safety and energy consumption across the entire hospital.

[0111] Example 8:

[0112] The HVAC control execution interface module is used to translate regular operating signals or optimal HVAC control commands into physical control signals of the underlying HVAC actuators and then send them down for execution.

[0113] This embodiment adds a necessary execution module to the system of Embodiment 1 to achieve a complete control closed loop. The system also includes an HVAC control execution interface module. The purpose of this module is to act as a translator and actuator between the upper-level algorithm's forward-looking risk hedging and energy consumption optimization module and the hospital's underlying building automation system (BAS) or HVAC physical actuators. In this embodiment, the HVAC control execution interface module monitors the output of the forward-looking risk hedging and energy consumption optimization module in real time. This output may be a normal operating signal, such as maintaining the status quo, or it may be the optimal HVAC control command calculated after triggering hedging. For example, a command vector containing {ICU damper opening +5%, corridor exhaust fan frequency +2%}; this module, for example, is a standardized interface gateway that implements building automation standard protocols such as BACnet and Modbus, responsible for transmitting these high-level logical commands. The module translates physical control signals into signals that the underlying HVAC actuators, such as fan inverters, electric dampers, and VAV terminal boxes, can understand, such as 0-10V analog voltage signals or switching signals; the module then sends these physical signals down to execute, driving the HVAC equipment to perform physical actions.

[0114] This embodiment, by adding this HVAC control execution interface module, bridges the final gap between upper-level intelligent algorithm decision-making and lower-level physical device execution, constructing a complete closed-loop control system. It ensures that the optimal HVAC control commands calculated by the proactive risk hedging and energy consumption optimization modules can be executed promptly and accurately, thereby truly realizing the control of pollutants during actual diffusion. The module proactively controls the environment beforehand; for example, HVAC operations result in new pressure differentials. It will be immediately re-collected by the multimodal data acquisition module at the next moment, forming a continuously running, self-correcting intelligent closed loop.

[0115] It should be noted that the above 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 preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A big data management and intelligent evaluation system for hospital environment air quality, characterized in that, The method comprises the following steps: A multi-modal data acquisition module is used to collect real-time environmental state data and dynamic crowd flow data; A data processing module is used to normalize real-time pressure difference data in the dynamic crowd flow data and the environmental state data, to obtain normalized crowd activity indicators and normalized real-time pressure difference; A dynamic transmission modeling module is used to combine a preset baseline aerodynamic coupling coefficient, the normalized crowd activity indicators, the normalized real-time pressure difference, and door state data in the dynamic crowd flow data, to solve a dynamic air transmission coefficient through a preset dynamic disturbance function; A dynamic Bayesian network risk modeling module is used to solve a cross-zone transmission risk probability according to the dynamic air transmission coefficient and pollutant concentration data in the environmental state data; A forward-looking risk hedging and energy consumption optimization module is used to compare the cross-zone transmission risk probability with a preset risk tolerance threshold, and to generate a normal operation signal in response to the cross-zone transmission risk probability being less than the risk tolerance threshold; In response to the cross-zone transmission risk probability being greater than or equal to the risk tolerance threshold, a target optimization problem is solved, which aims to minimize a preset energy consumption cost function and is constrained by a simulated risk probability being lower than a preset constraint value, to solve optimal HVAC control instructions; The cross-zone transmission risk probability is defined as a conditional probability that, when a specific pollutant event occurs in a source zone and under current observation evidence, the pollutant concentration in a target critical zone exceeds a preset cleanliness threshold within a future time step.

2. The system of claim 1, wherein, The environmental state data includes particulate matter concentration, microorganism concentration, carbon dioxide concentration, temperature and humidity, and differential pressure data between critical zones.

3. The system of claim 1, wherein, The dynamic crowd flow data includes crowd density, flow direction, average residence time, and door state data.

4. The system of claim 1, wherein, The preset baseline aerodynamic coupling coefficient is calibrated by an aerodynamic atlas construction module; wherein the aerodynamic atlas construction module defines hospital functional areas as nodes and inter-regional physical connections as edges, and calibrates the baseline aerodynamic coupling coefficient through computational fluid dynamics simulation or tracer gas experiments.

5. The system of claim 1, wherein, The preset constraint value is the product of the preset risk tolerance threshold and a preset safety redundancy coefficient.

6. The system of claim 1, wherein, The simulated risk probability is calculated through "What-if” deduction; The "What-if” deduction comprises: Based on the HVAC control instruction vector to be evaluated and the current pressure difference data, a new pressure difference after regulation and control is predicted through an HVAC-pressure response model; The new pressure difference is substituted into the preset dynamic disturbance function to calculate a simulated dynamic transmission coefficient; The simulated dynamic transmission coefficient is used, and based on the current state, Bayesian network forward inference is re-executed to calculate the simulated risk probability.

7. The system of claim 1, wherein, The preset risk tolerance threshold is differentiated according to the medical function importance of the target zone and the cleanliness level requirement.

8. The system of claim 1, wherein, Further comprising: An HVAC control execution interface module is used to translate the normal operation signal or the optimal HVAC control instruction into a physical control signal of a bottom-layer HVAC executor, and to issue and execute.

Images