Comprehensive air environment adjusting system for hospital purification area

Abstract

The invention discloses an air environment comprehensive regulation system for a hospital purification area, and the system comprises S1, an environment parameter sensing node network which is used for obtaining data and generating sensing node data; s2, an area function identification and state labeling module which is used for generating an area state description vector; s3, a multi-modal environment state fusion module which is used for constructing an environment dynamic state map of the current area and extracting key influence factors; s4, an adaptive fuzzy control decision module generates an air conditioning parameter set through fuzzy logic and a dynamic weight adjustment mechanism; s5, the multi-channel air conditioning execution device is used for adjusting the regional air according to the air conditioning parameter set; and S6, a comfort level feedback and strategy optimization module which is used for performing closed-loop optimization on the fuzzy rule set in the adaptive fuzzy control decision module and the weight factor of each rule. The method has the advantages of being fine in response granularity, adaptive in control strategy and high in human factor comfort degree fusion degree.

Description

Technical Field

[0001] The present invention relates to the technical field of environmental regulation, and particularly to an integrated air environment regulation system for hospital purification areas. Background Art

[0002] With the continuous improvement of the construction standards for precision medical environments, higher requirements have been put forward for air cleanliness, temperature and humidity control, and oxygen supply safety in hospital purification areas. In multi-functional composite medical spaces, how to dynamically adapt to the air environment requirements under different usage states on the basis of meeting environmental control specifications has become a key problem faced by modern hospital building environment control systems.

[0003] In the prior art, the air conditioning systems in hospital purification areas mainly rely on centralized air conditioning and ventilation systems. The control methods generally adjust the air supply parameters based on fixed-time strategies or local temperature and humidity detection results. The core algorithms are mainly based on preset control logics or static PIDs, lacking the ability to comprehensively judge the actual regional usage states, human factor comfort parameters, and multi-source environmental variables. In addition, most of the existing systems lack real-time perception and response mechanisms for the behavior density of medical staff, the physiological states of patients, and the dynamic changes in air quality, resulting in problems such as adjustment lags, rough responses, and insufficient comfort adaptation in different purification areas, making it difficult to meet the dual requirements of personalized control and energy-saving efficiency in medical scenarios.

[0004] Currently, some studies have begun to attempt to introduce intelligent means such as fuzzy control and Internet of Things perception, but they are still generally limited to a single control dimension (such as temperature and humidity) or simple condition-triggered models, failing to achieve a comprehensive regulation mechanism for multi-source perception input, state modeling output, and dynamic feedback closed-loop. Moreover, a direct coupling relationship between the air conditioning results and user comfort has not been established. The existing control systems lack the ability of graph model structures to jointly model the airflows, physical adjacencies, and functional states between different purification areas, and also do not construct a path to input the graph structure information into the control decision-making system, resulting in weak inter-regional collaborative regulation ability and rough control granularity. In addition, in terms of the comfort feedback mechanism, most of the existing methods use empirical threshold adjustments, lacking the ability to comprehensively model subjective scores and physiological parameters, and also unable to support the evolution of control strategies with dynamically updated rule set weights.

[0005] In summary, there is an urgent need for an air environment regulation system that integrates multi-source environmental perception, spatial graph modeling, fuzzy inference decision-making, and human factor feedback optimization to solve the core problems such as untimely air conditioning responses, rigid control strategies, insufficient regional adaptability, and lack of comfort feedback paths in current hospital purification areas. Summary of the Invention

[0006] An object of the present invention is to propose an integrated air environment regulation system for hospital purification areas. The present invention integrates multi-parameter environmental perception, regional state modeling, and intelligent control feedback mechanisms, constructs a regional state atlas and an adaptive fuzzy control decision-making model based on a graph convolutional network, and realizes closed-loop optimization control of air conditioning parameters by driving the iterative update of fuzzy rule weights through a comfort feedback matrix, with the advantages of fine response granularity, adaptive control strategies, and high integration of human factor comfort.

[0007] According to the integrated air environment regulation system for hospital purification areas of an embodiment of the present invention, the system includes: S1. An environmental parameter perception node network, which is arranged in different cleanliness level areas within the purification area and is equipped with temperature and humidity sensors, oxygen concentration sensors, carbon dioxide concentration sensors, and air suspension particle concentration sensors to generate perception node data; S2. A regional function identification and status annotation module, which is used to obtain the activity information of medical staff in each area, the in-bed status of patients, and the operation data of medical equipment, and generate a regional state description vector; S3. A multi-modal environmental state fusion module, which is used to perform feature mapping on the perception node data and the regional state description vector, construct an environmental dynamic state atlas of the current area, and extract key influencing factors through a graph convolution method; S4. An adaptive fuzzy control decision-making module, which generates a set of air conditioning parameters based on the key influencing factors of the environmental dynamic state atlas through a fuzzy logic and dynamic weight adjustment mechanism; S5. A multi-channel air conditioning execution device, which is used to adjust the air volume, supply air temperature, fresh air ratio, and oxygen supply rate of the area respectively according to the set of air conditioning parameters; S6. A comfort feedback and strategy optimization module, which is used to periodically collect the physiological parameters and subjective comfort scores of medical staff and patients in the area and generate a comfort feedback matrix, and perform closed-loop optimization on the fuzzy rule set and the weight factors of each rule in the adaptive fuzzy control decision-making module based on the comfort feedback matrix.

[0008] Optionally, the S2 specifically includes: S21. A personnel activity recognition unit, which is used to collect the positioning trajectories, residence durations, and movement patterns of medical staff in each area; S22. An activity frequency calculation sub-module, which is used to preprocess the positioning trajectories, residence durations, and movement patterns obtained by the personnel activity recognition unit, extract the number of activities, residence duration, and activity hot spot distribution per unit time, and calculate the regional activity frequency parameter :

[0009] Wherein, represents the The residence duration of medical staff in this area, indicating the number of activities of the th medical staff, indicating the activity hot zone activation index of this area within the time interval being a preset weight coefficient, indicating the number of medical staff present in the said area within the observation time interval ; S23. Patient status acquisition unit, which is used to collect information on whether the patient is in bed and whether the patient is in a resting state through the bed management system interface and the bed sensor, and obtain the bed occupancy rate, and form patient status parameters , where , indicating the number of in-bed patients, indicating the total number of beds in the area; S24. Equipment operation status identification unit, the equipment operation status identification unit includes data interface modules for air purification equipment, medical gas equipment and negative pressure devices, and is used to collect the on / off status, operating power and alarm information of various equipment, and convert them into equipment status vectors ; S25. Area usage level determination module, which is used to combine the , and three groups of parameters, and classify the area function status by hierarchical weighted scoring; S26. Status vector generation module, which is used to perform data standardization and feature fusion on the area activity frequency parameters, patient status parameters and equipment status vectors according to rules, and construct an area status description vector .

[0010] Optionally, the S3 specifically includes: S31. Input data receiving unit, which is used to receive perception node data and area status description vectors; S32. Data standardization processing sub-module, which is used to perform normalization processing on the perception node data and area status description vectors, convert different physical quantities into vector data representations within a unified dimension range, and generate a fusion input matrix; S33. Environment map construction module, which is used to construct an undirected weighted graph structure based on the fusion input matrix, in combination with the physical space adjacency relationship, ventilation air flow path and functional relevance between each purification sub-area; S34. Graph convolution feature extraction sub-module, which is used to calculate the status feature vector of each area node based on the undirected weighted graph structure and the fusion input matrix through a graph convolution network model :

[0011] Among them, represents the set of adjacent nodes to the regional node ; is the regional node and the regional node is the edge weight value between them; represents the fusion input matrix corresponding to the regional node ; is the weight parameter matrix in the graph convolutional network; is the activation function; S35, the key influencing factor screening unit, is used to identify the key influencing factors affecting the current regional air environment state based on the extracted state feature vector, using the threshold discrimination and change trend analysis method.

[0012] Optionally, the S35 specifically includes: S351, the feature change monitoring module, is used to perform sliding window sampling on the feature values of each dimension in the state feature vector to construct a historical sequence set , where represents the th feature value at the current moment ; is the window length; S352, the threshold discrimination unit, is used to calculate the deviation between each feature value and its historical mean to obtain the change amplitude index . If exceeds the preset sensitivity threshold , then mark this feature as a candidate influencing factor; S353, the change trend scoring module, is used to calculate the linear fitting slope based on the time series trend in the historical sequence set and compare it with the system-set trend threshold . If , then regard the corresponding feature as a candidate factor with significant trend; S354, the joint screening decision module, is used to perform an intersection operation on the feature indicators that satisfy and to generate a key influencing factor set , where represents the th key influencing factor obtained through discrimination and screening.

[0013] Optionally, S4 specifically includes: S41. A fuzzy rule base construction unit, configured to define a set of fuzzy input variables and an output variable set according to a preset purification environment management policy and establish a set of input-output fuzzy mapping rules ; S42. A fuzzy membership degree calculation module, configured to map each eigenvalue in the set of key influencing factors to the membership function values of the set of fuzzy input variables , and calculate the fuzzy value matrix of the input variables by using a triangular membership function ; S43. An inference engine module, configured to perform fuzzy logic inference according to the current fuzzy value matrix and the set of fuzzy mapping rules , and output a fuzzy inference result ; S44. A dynamic weight adjustment sub-module, configured to adjust the weight factors of each rule in the set of fuzzy mapping rules based on historical control results and current regional comfort feedback information , and form a weighted fuzzy rule set , where represents the th fuzzy rule, represents the updated weight value corresponding to this rule, is the total number of rules in the set of fuzzy rules; S45. A defuzzification module, configured to perform defuzzification processing on the fuzzy inference result by using the centroid method to generate a set of air conditioning parameters .

[0014] Optionally, S5 specifically includes: S61. A physiological parameter acquisition unit, configured to acquire the physiological parameters of medical staff and patients in the area through wearable and non-contact physiological sensing devices, where the physiological parameters include skin temperature, heart rate, respiratory rate, and skin conductivity, and form a physiological state vector , where represents the real-time measured value of the th physiological parameter; S62. A subjective comfort score input unit, configured to acquire the subjective comfort score of users on the current air environment based on a mobile terminal, a panel device, and a voice system, and the score value corresponding to the subjective comfort score forms a subjective evaluation vector , where represents the score value of the rd user on the environment in the current cycle; S63, Comfort Feedback Matrix Generation Module, for weighting and fusing the physiological state vector and the subjective evaluation vector to generate a comfort feedback matrix , where the comfort feedback matrix represents the comprehensive comfort feedback degree of individual users to each environmental control variable:

[0015] Among them, is the matrix expression after normalization of the physiological state vector , is the matrix expression after normalization of the subjective evaluation vector , is the fusion weight coefficient, representing the relative weight of physiological and subjective scores in the comprehensive evaluation, represents the comfort feedback value of the th type of user to the th air control parameter; S64, Policy Feedback Optimization Unit, for comparing the comfort feedback matrix with the historical control output results of the fuzzy control system, constructing an objective error function, and based on the objective error function, using the backpropagation mechanism to update and adjust the weight factors of each rule in the fuzzy rule set , where represents the updated weight factor of the th rule after optimization, represents the th fuzzy rule, is the total number of rules in the fuzzy rule set.

[0016] Optionally, the S64 specifically includes: S641, Historical Output Data Caching Module, for periodically storing the air conditioning parameter set , where represents the actual output value of the th air conditioning parameter in the current cycle ; S642, Comfort Target Mapping Module, for converting the user feedback data in the comfort feedback matrix into a target adjustment value set , where represents the set value of the 2nd air conditioning parameter expected by the user; S643, Error Function Construction Unit, for defining an objective error function based on the difference between the current cycle adjustment output value and the user expected value; S644, a weight update module, is used to adjust the weight factors of each rule in the fuzzy rule set according to the backpropagation update mechanism based on the target error function :

[0017] wherein represents the learning rate, is the weight factor of the -th fuzzy rule in the current cycle, are the updated weight factors of each rule, represents the gradient of the error with respect to the weight of this rule; S645, an optimized rule set output unit, is used to apply the updated rule weights to the next round of fuzzy control inference process to construct an updated weighted fuzzy rule set .

[0018] The beneficial effects of the present invention are as follows: (1) The present invention constructs a multi-source environment perception system for the hospital purification area, deploys high-precision temperature and humidity, gas concentration and particle concentration sensing nodes in the main activity areas of medical staff and patients, and introduces a regional usage status modeling mechanism. By fusing the frequency of personnel activities, the occupancy status of patient beds and the operation characteristics of equipment, a multi-dimensional regional description vector is formed, providing a structured state input for subsequent intelligent regulation, and significantly improving the perception dimension and parameter integrity of the air conditioning system for actual space usage changes.

[0019] (2) The present invention proposes a method for constructing an environmental dynamic state map based on a graph convolutional network, spatially adjacently fuses the sensing node data with the regional functional status, forms an undirected weighted graph structure covering the air flow path, functional partition and physical connection relationship, and extracts key air state influencing factors based on node feature propagation, establishing a new path for internal regional-level air state evolution modeling in the purification area, and significantly enhancing the spatial relevance and regional adaptability of the regulation strategy.

[0020] (3) The present invention designs an adaptive fuzzy control decision-making mechanism, constructs fuzzy input variables by combining the key influencing factors extracted from the state map, generates a control parameter set through dynamic adjustment based on the membership function and weight rules, and drives the periodic update of the fuzzy rule set weights through a comfort feedback matrix, constructing a control rule self-evolution mechanism for actual use experience, breaking the problems of fixed traditional control logic and lagging feedback, and improving the flexibility of regulation response and individual adaptability.

[0021] ​(4) The present invention introduces a comfort feedback matrix fusion method, which unifies the modeling of physiological parameters and subjective scores, quantifies the comprehensive feedback of users under various air control variables through a matrix structure, and optimizes the output of the control system in combination with the error backpropagation mechanism. The system can achieve a closed-loop iterative control path of human factor perception - state response - rule reconstruction during the dynamic adjustment process, greatly improving the personalized adjustment ability of the system and the control accuracy driven by comfort. Description of the Drawings

[0022] The drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation to the present invention. In the drawings: Figure 1 It is the overall structural framework diagram of the comprehensive air environment adjustment system for the hospital purification area proposed by the present invention. Detailed Embodiment

[0023] Now, the present invention will be further described in detail with reference to the drawings. These drawings are all simplified schematic diagrams, only showing the basic structure of the present invention in a schematic way, so they only show the components related to the present invention.

[0024] Reference Figure 1 , for the comprehensive air environment adjustment system in the hospital purification area, the system includes: S1. An environmental parameter perception node network, which is arranged in different cleanliness level areas in the purification area and is equipped with temperature and humidity sensors, oxygen concentration sensors, carbon dioxide concentration sensors, and air suspension particle concentration sensors to generate perception node data; In this embodiment, the environmental parameter perception node network is distributed in different cleanliness level areas in the hospital purification area to form a perception unit layout with balanced spatial coverage. Each perception node integrates a temperature and humidity sensor, an oxygen concentration sensor, a carbon dioxide concentration sensor, and a laser particle sensor, which are used to synchronously collect the air physical and gas component parameters of the target area, including but not limited to physical indicators such as temperature, relative humidity, wind speed, and particle concentration, as well as gas component parameters such as oxygen concentration, carbon dioxide concentration, volatile organic compounds, and carbon monoxide concentration, so as to comprehensively describe the air quality status of each functional area in the hospital purification area. The sensing nodes use wired or wireless communication methods to transmit data to the system main control platform. Each node has a unique address identifier and supports data annotation based on time stamps. The system controller periodically collects the sensing data uploaded by each node to form a perception node data set containing spatial coordinates, time information, and environmental indicators for subsequent environmental state modeling and control strategy input. The installation positions of the nodes are optimized and differentially distributed according to the regional functional attributes such as air supply outlets, air return outlets, hospital bed areas, and equipment areas to ensure comprehensive coverage and dynamic response capabilities for the air environment state.

[0025] S2, the area function recognition and status annotation module, is used to obtain the activity information of medical staff in each area, the in-bed status of patients, and the operation data of medical equipment, and generate an area status description vector; In this embodiment, S2 specifically includes: S21, the personnel activity recognition unit, is used to collect the positioning trajectories, residence durations, and movement patterns of medical staff in each area; S22, the activity frequency calculation sub-module, is used to preprocess the positioning trajectories, residence durations, and movement patterns obtained by the personnel activity recognition unit, extract the number of activities, residence duration, and activity hot zone distribution per unit time, and calculate the area activity frequency parameter :

[0026] Among them, represents the residence duration of the th medical staff in this area, represents the number of activities of the th medical staff, represents the activity hot zone activation index of this area in the time interval , is a preset weight coefficient, represents the observation time interval and represents the number of medical staff appearing in this area within; This formula is used to quantify the overall activity level of personnel activities in a certain purification area per unit time. This frequency parameter is composed of three parts weighted: the first part is the average residence duration of all personnel in this area , reflecting the density of personnel residence; the second part is the average number of activities of personnel , indicating the frequency of behavior per unit time; the third part is the activity hot zone activation index in the area , that is, the activity level of high-frequency activity areas in the spatial dimension. The weight coefficient corresponds to the adjustment factors of the above three indicators respectively, and is used to dynamically adjust the influence weights of various factors on the final activity frequency so as to realize the comprehensive evaluation of the flow intensity and spatial distribution characteristics of the flow of people in different purification areas at different time periods. This parameter is finally used as an important input for area function recognition to guide the generation of intelligent control strategies for the air environment.

[0027] S23, the patient status collection unit, is used to collect information on whether the patient is in bed and whether the patient is in a resting state through the bed management system interface and bed sensors, and obtain the bed occupancy rate to form patient status parameters , where , represents the number of in-bed patients, and represents the total number of beds in the area; S24. Equipment operation status recognition unit. The equipment operation status recognition unit includes data interface modules for air purification equipment, medical gas equipment, and negative pressure devices, and is used to collect the switch status, operating power, and alarm information of various equipment, and convert them into equipment status vectors ; S25. Area usage level determination module, which is used to combine the , and three groups of parameters, and classify the functional status of the area by hierarchical weighted scoring; S26. Status vector generation module, which is used to perform data standardization and feature fusion on the area activity frequency parameter, patient status parameter, and equipment status vector according to rules, and construct an area status description vector .

[0028] In this embodiment, multi-source heterogeneous data such as the positioning trajectories of medical staff, activity frequencies, the occupancy of patient beds, and the operation status of medical equipment are integrated to construct an area function recognition and status annotation module. By structurally encoding spatial behavior information and functional usage, an area status description vector is generated, which not only reflects the current usage intensity of the area but also reflects its dynamic evolution trend, realizing the refined recognition of different functional spaces in the purification area and the modeling of the usage status, and improving the system's accurate matching ability for local air conditioning requirements.

[0029] The multi-modal environment status fusion module is used to perform feature mapping on the perception node data and the area status description vector, construct an environmental dynamic status map of the current area, and extract key influencing factors through a graph convolution method; In this embodiment, S3 specifically includes: S31. Input data receiving unit, which is used to receive perception node data and the area status description vector; S32. Data standardization processing sub-module, which is used to perform normalization processing on the perception node data and the area status description vector, convert different physical quantities into vector data representations within a unified dimension range, and generate a fusion input matrix; S33. Environmental map construction module, which is used to construct an undirected weighted graph structure based on the fusion input matrix, combined with the physical space adjacency relationship, ventilation air flow path, and functional relevance between each purification sub-area; S34. Graph convolution feature extraction sub-module, which is based on the undirected weighted graph structure and the fusion input matrix , and calculates the status feature vector of each area node through a graph convolution network model :

[0030] Among them, represents the set of adjacent nodes to the regional node adjacent node set, is the regional node and the regional node the edge weight value between them, represents the corresponding fusion input matrix at the regional node location, is the weight parameter matrix in the graph convolutional network, is the activation function; This formula is used to describe the state feature vector of the regional node in the graph convolutional feature extraction sub-module calculation process. This formula is based on the basic propagation mechanism of the graph convolutional network and performs a weighted aggregation operation on the features of adjacent nodes for each node in the undirected weighted graph structure. Specifically, represents the set of nodes adjacent to node adjacent node set, is the node and the adjacent node the edge weight value between them, reflecting the connection strength between nodes; is the node in the fusion input matrix, is the trainable weight matrix in the graph convolutional model, used for linear transformation of node features; is the activation function, used to introduce non-linear expressive ability.

[0031] S35, the key influencing factor screening unit, is used to identify the key influencing factors affecting the current regional air environment state based on the extracted state feature vector, using the threshold discrimination and change trend analysis method.

[0032] The said S35 specifically includes: S351, the feature change monitoring module, is used to perform sliding window sampling on the feature values of each dimension in the state feature vector to construct a historical sequence set , where represents the th feature value at the current time , is the window length; S352, the threshold discrimination unit, is used to calculate the deviation between each feature value and its historical mean to obtain the change amplitude index , if exceeds the preset sensitivity threshold , then mark this feature as a candidate impact factor; S353. A change trend scoring module for calculating the linear fitting slope based on the time series trend in the historical sequence set and comparing it with the trend threshold set by the system , and if , then regard the corresponding feature as a candidate factor with significant trend; S354. A joint screening decision module for performing an intersection operation on the feature indicators that meet and to generate a set of key impact factors , where represents the th key impact factor obtained through discriminant screening.

[0033] In this embodiment, by jointly feature mapping the multi-dimensional air quality data collected by the environmental perception nodes and the regional function state description vector, an environmental dynamic state map with clear spatial relationships and rich parameter dimensions is constructed. A graph convolutional network is introduced to model the feature propagation relationships between regional nodes, which not only retains the structural connection information between different sub-regions in terms of air flow paths, physical adjacency, and functional coupling, but also extracts the key impact factors in the high-dimensional state data, realizing the air environment modeling path from local perception to global association, improving the structural understanding and differential control capabilities of the air conditioning system for complex spatial states, and providing a deep data support and spatial perception basis for achieving precise and efficient air quality regulation.

[0034] An adaptive fuzzy control decision module generates a set of air conditioning parameters based on the key impact factors of the environmental dynamic state map through a fuzzy logic and dynamic weight adjustment mechanism; In this embodiment, S4 specifically includes: S41. A fuzzy rule base construction unit for defining a set of fuzzy input variables and a set of output variables according to the preset purification environment management strategy, and establishing a set of input-output fuzzy mapping rules ; S42. A fuzzy membership degree calculation module for mapping each eigenvalue in the set of key impact factors to the membership function values of the set of fuzzy input variables , and calculating the fuzzy value matrix of the input variables using a triangular membership function; S43. An inference engine module for, according to the current fuzzy value matrix and the set of fuzzy mapping rules ​, perform fuzzy logic reasoning and output the fuzzy reasoning result ; S44. The dynamic weight adjustment sub-module is used to adjust the weight factors of each rule in the fuzzy mapping rule set based on the historical control results and the current regional comfort feedback information , and form a weighted fuzzy rule set , where represents the -th fuzzy rule, represents the updated weight value corresponding to this rule, is the total number of rules in the fuzzy rule set; S45. The defuzzification module is used to perform defuzzification processing on the fuzzy reasoning result by the centroid method to generate the air conditioning parameter set .

[0035] Based on the key influencing factors extracted from the environmental dynamic state atlas, this embodiment constructs the fuzzy control input, combines the fuzzy logic reasoning mechanism with the dynamic weight adjustment strategy, realizes the real-time generation and continuous optimization of the air conditioning parameters. The system can automatically match the fuzzy rule set according to the actual regional state and update the weights. It not only has the flexible control ability to process multi-dimensional uncertain information, but also realizes the adaptive iteration of the control rules through the comfort feedback mechanism, thus enhancing the intelligent responsiveness and personalized adaptation level of the air environment control process, and providing stable and flexible intelligent control support for the dynamic air management in complex medical spaces.

[0036] S5. The multi-channel air conditioning execution device is used to adjust the air volume, supply air temperature, fresh air ratio and oxygen supply rate of the area according to the air conditioning parameter set respectively; In this embodiment, the multi-channel air conditioning execution device includes a supply air control unit, a temperature adjustment unit, a fresh air mixing module and a medical oxygen supply allocation module. Each sub-module corresponds to the air volume, supply air temperature, fresh air ratio and oxygen supply rate in the air conditioning parameter set respectively. The supply air control unit adjusts the air volume output of each purification sub-area through a variable air volume damper; the temperature adjustment unit uses an electric cooling and heating coil valve to adjust the supply air temperature in real time; the fresh air mixing module dynamically matches according to the set fresh air ratio through a fresh air and return air ratio control valve group; the medical oxygen supply allocation module is connected to the hospital central oxygen supply system and adjusts the oxygen flow through a proportional electric control valve. A real-time communication channel is established between each module and the central control system, and independent execution is carried out according to the corresponding parameters in the generated air conditioning parameter set, and the output is corrected and adjusted in a closed loop by combining the feedback sensor data, so as to realize the differential and precise control of the air environment parameters in different areas.

[0037] A comfort feedback and strategy optimization module, which is used to periodically collect the physiological parameters and subjective comfort scores of medical staff and patients in the area, generate a comfort feedback matrix, and perform closed-loop optimization on the fuzzy rule set and the weight factors of each rule in the adaptive fuzzy control decision module based on the comfort feedback matrix.

[0038] In this embodiment, S6 specifically includes: S61. A physiological parameter acquisition unit, which is used to collect the physiological parameters of medical staff and patients in the area through wearable and non-contact physiological sensing devices. The physiological parameters include skin temperature, heart rate, respiratory rate, and skin conductivity, and form a physiological state vector , where represents the real-time measured value of the th physiological parameter; S62. A subjective comfort score input unit, which is used to collect the subjective comfort scores of users on the current air environment based on mobile terminals, panel devices, and voice systems. The score values corresponding to the subjective comfort scores are linearly normalized to form a subjective evaluation vector , where represents the score value of the th user on the environment in the current period; S63. A comfort feedback matrix generation module, which is used to perform weighted fusion on the physiological state vector and the subjective evaluation vector to generate a comfort feedback matrix . The comfort feedback matrix represents the comprehensive comfort feedback degree of individual users on each environmental control variable:

[0039] where is the matrix expression of the normalized physiological state vector , is the matrix expression of the normalized subjective evaluation vector , is the fusion weight coefficient, representing the relative weight of physiological and subjective scores in the comprehensive evaluation, represents the th type of user's comfort feedback value on the th air control parameter; This formula is used to describe the fusion process of the subjective and objective evaluation data in the comfort feedback matrix generation module. This module fuses the normalized physiological state vector matrix and the subjective score vector matrix in a weighted manner to construct a comfort feedback matrix , ​​​​is an adjustable fusion weight coefficient, which is used to control the relative contribution of physiological evaluation and subjective score in the overall comfort feedback; represents the category of users' comfort feedback value for the th air control parameter. This matrix realizes the unified expression of multi-user, multi-parameter, and multi-source data, providing a quantitative basis and data foundation for subsequent control strategy optimization and personalized air conditioning.

[0040] S64, the policy feedback optimization unit, is used to compare the comfort feedback matrix with the historical control output results of the fuzzy control system, construct an objective error function, and update and adjust the weight factors of each rule in the fuzzy rule set based on the objective error function using the backpropagation mechanism to update and form a new weighted fuzzy rule set , where represents the updated weight factor of the th rule after optimization, represents the th fuzzy rule, is the total number of rules in the fuzzy rule set.

[0041] The specific content of the S64 includes: S641, the historical output data caching module, is used to periodically store the air conditioning parameter set , where represents the actual output value of the th air conditioning parameter in the current cycle ; S642, the comfort target mapping module, is used to convert the user feedback data in the comfort feedback matrix into the target adjustment value set , where represents the set value of the 2nd air conditioning parameter expected by the user; S643, the error function construction unit, is used to define the objective error function based on the difference between the current cycle adjustment output value and the user expected value as:

[0042] where, represents the total number of air conditioning parameters, represents the current cycle adjustment output value of the th air conditioning parameter, represents the set value of the th air conditioning parameter expected by the user; S644. A weight update module for adjusting the weight factors of the rules in the fuzzy rule set according to the backpropagation update mechanism based on the target error function : where

[0043] denotes the learning rate, is the weight factor of the -th fuzzy rule in the current cycle, is the weight factor of each updated rule, represents the gradient of the error with respect to the weight of this rule; This formula is used to describe the update mechanism of the weight factors of the fuzzy control rule set in the weight update module. Based on the value of the target error function , the gradient descent method is used to adjust the weight factor of the -th fuzzy rule in the current cycle, where represents the weight factor of the -th fuzzy rule in the current cycle, represents the updated weight factor in the next cycle, is the learning rate used to control the step size of each weight adjustment, is the partial derivative of the error function with respect to the weight of this rule, that is, the influence degree of the current rule weight on the overall regulation error of the system. By minimizing the error function, this formula realizes the dynamic weight adjustment of each rule in the fuzzy control rule set under multi-cycle feedback, thereby improving the adaptive ability and control accuracy of the regulation strategy.

[0044] S645. An optimized rule set output unit for applying the updated rule weights to the next round of fuzzy control inference process to construct an updated weighted fuzzy rule set .

[0045] In this embodiment, by integrating the physiological parameters and subjective comfort scores of medical staff and patients, a comfort feedback matrix is constructed to quantify the comprehensive perception feedback of different user groups on various air control parameters, and based on this, a target error function is constructed. By introducing the backpropagation mechanism to periodically optimize the weight factors in the fuzzy control rule set, a closed-loop adjustment path from perception - evaluation - control - re-optimization is realized, which not only reflects the difference between the current air conditioning effect and the user's expectation, but also promotes the continuous learning and dynamic adaptation of the control strategy, significantly improving the personalized regulation ability and comfort matching accuracy of the system in multi-state scenarios.

[0046] Example: To verify the applicability and effectiveness of the system of the present invention in the actual medical environment, the present invention was deployed in the purification ward of the newly built inpatient department of a Class III Grade A general hospital in East China. The floor area of this ward is about 3,800 square meters, and it has multiple purification areas such as high-level clean operating rooms, ICU wards, infection control buffer rooms, and isolation observation areas. The regional functions are complex and the environmental control requirements are high. It is a typical scenario where the traditional air conditioning system is most likely to have a lag in adjustment response and a mismatch in comfort. Previously, a centralized variable air volume (VAV) system combined with the traditional PID control method was adopted in this ward, lacking the adjustment ability based on the actual flow state of people and comfort feedback. Especially during the peak period of medical staff handover, low-power operation of equipment at night, and intensive periods of patient transfer, the fluctuations of regional air temperature, humidity, and carbon dioxide concentration are large, and phenomena such as environmental adjustment lag, uneven air supply, and too low oxygen content in the air often occur, seriously affecting the medical staff efficiency and patient comfort.

[0047] The present invention was pilot-deployed in this ward. First, an environmental parameter perception node network was respectively arranged in 12 core functional sub-regions including the ICU, the surgical area, and the isolation observation area. The total number of nodes is 36. Each node integrates a temperature and humidity sensor, a carbon dioxide concentration probe, an oxygen concentration module, and a laser particle sensor to ensure that each region has at least 3 equally spaced perception points to achieve dynamic coverage of the local air state. The system collects the positioning trajectories, activity frequencies, and residence durations of medical staff in real time through the combination of positioning tags and video analysis technology, and converges the patient bed sensing data (such as in-bed state, resting state, abnormal heart rate) and equipment operation states (such as medical gas valve state, air purification equipment alarm information) to form a regional state description vector. An environmental dynamic state map is constructed through a multi-modal fusion method, and the key influencing factors in each time period are extracted by a graph convolutional network as the input of intelligent control.

[0048] In the process of generating the control strategy, the system adopts a fuzzy control rule set, combines the extracted key factors (such as the increase in people flow density, frequent changes in active hot zones, and the decrease in patient oxygen saturation, etc.), and outputs a set of adjustment parameters through a fuzzy inference mechanism, including the fresh air ratio, supply air temperature, air volume adjustment coefficient, and oxygen supply rate. At the same time, the system forms a multi-dimensional comfort feedback matrix by collecting the comfort scores and physiological indicators (skin temperature, respiratory rate, heart rate variability, etc.) of medical staff and patients, and combines the error function construction method to realize the dynamic learning and update of the weight parameters in the fuzzy control rules, so that the adjustment strategy has the adaptive optimization ability of "human factor feedback drive".

[0049] The operation cycle after the system deployment is 30 days. A systematic evaluation was carried out on the changes in air environment indicators, comfort levels, and energy use efficiency before and after the deployment. The results are shown in Table 1 and Table 2.

[0050] Comparison Report on Key Air Environment Indicators in the Ward Area 1 Monitoring indicators Daily average value before deployment Daily average value after deployment MoM change Stability improvement rate Temperature fluctuation range (°C) 2.3 0.7 -69.6% +71.2% Humidity fluctuation range (%) 7.5 2.4 -68.0% +68.9% <![CDATA[CO 2 Peak concentration (ppm)]]> 1430 875 -38.8% +55.4% <![CDATA[O 2 Lowest concentration (%)]]> 19.1 20.4 +6.8% +53.1% Number of instantaneous PM2.5 exceedance times (times) 18 3 -83.3% +86.5% Comparison Report on Comfort Indexes of Medical Staff and Patients 2 Evaluation dimension Score before deployment (on a 5-point scale) Score after deployment (on a 5-point scale) Increase rate Data sample Comprehensive score of medical staff 3.6 4.7 +30.5% 146 copies Patient subjective comfort score 3.9 4.8 +23.1% 84 copies Proportion of heart rate stable interval (%) 63.2 81.5 +28.9% 128 groups Number of abnormal respiratory rhythms (times) 74 21 -71.6% 96 groups Sleep interruption rate (%) 18.4 7.2 -60.9% 91 cases The data results show that after the deployment of the system of the present invention, the volatility of the regional air environment has decreased significantly. Especially in the ICU area, due to the high proportion of patients using ventilators and being extremely sensitive to changes in oxygen concentration, the system can adjust the oxygen supply rate immediately when detecting fluctuations in oxygen concentration, reducing the risk of hypoxia. The subjective scoring data shows that medical staff generally feedback that "the air is fresher", "it is not easy to feel dizzy during the night shift", "the operation area is quieter and drier", and the number of times of interrupted sleep and the frequency of abnormal heart rate of patients at night have also decreased significantly. The system continuously corrects the control strategy through the comfort feedback matrix to make the air conditioning more in line with human factors requirements.

[0051] It is particularly worth mentioning that the system once captured a sudden oxygen supply fluctuation in a certain area of the ICU on the 18th day, and immediately linked to adjust the fresh air ratio and the oxygen supply valve, and restored the oxygen concentration to the normal value range within only 8 minutes. Compared with the average response time of 21 minutes for manual intervention required by the traditional system, the response efficiency of this system has increased by nearly 160%. In addition, the energy efficiency ratio of the system has increased significantly. According to the statistics of the logistics department, the daily average energy consumption of the air conditioning system in the deployment area has decreased by about 12.3%, achieving obvious energy-saving effects on the premise of ensuring air quality and comfort.

[0052] This embodiment shows that the comprehensive air environment regulation system proposed by the present invention has high practicability in the complex scenarios of the hospital purification area, can effectively solve the problems that the traditional system cannot dynamically perceive the regional state, the regulation strategy is static and rigid, and the lack of human factor feedback mechanism, significantly improves the response adaptability, comfort matching degree and control accuracy of the air environment, and provides an intelligent, high-efficiency and closed-loop optimized air environment control technical path for highly sensitive spaces in medical buildings.

[0053] The above is only the preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and all should be covered by the protection scope of the present invention.

Claims

1. The comprehensive air environment conditioning system of the hospital purification area is characterized by: The system comprises: S1, Environmental parameter sensing node network, which is deployed in different cleanliness level areas in the purification area and equipped with temperature and humidity, oxygen concentration, carbon dioxide concentration and airborne particle concentration sensors to generate sensing node data; S2, regional function identification and status labeling module, used to obtain the medical staff activity information, patient bed status and medical equipment operation data in each area, and generate regional status description vector; S3, multimodal environment state fusion module, is used to perform feature mapping between the perception node data and the regional state description vector, build the environmental dynamic state map of the current area, and extract key influencing factors through the graph convolution method; S4, adaptive fuzzy control decision module, based on the key influencing factors of the environmental dynamic state map, generates a set of air conditioning parameters through fuzzy logic and dynamic weight adjustment mechanism; S5, a multi-channel air conditioning execution device, used to adjust the air volume, air supply temperature, fresh air ratio and oxygen supply rate of the area according to the air conditioning parameter set; S6, comfort feedback and strategy optimization module, is used to periodically collect the physiological parameters and subjective comfort scores of medical staff and patients in the area and generate a comfort feedback matrix, and based on the comfort feedback matrix, perform closed-loop optimization on the fuzzy rule set and the weight factors of each rule in the adaptive fuzzy control decision module.

2. The comprehensive air environment conditioning system for the hospital purification area according to claim 1 is characterized in that: The S2 specifically includes: S21, personnel activity identification unit, used to collect the location trajectory, residence time and movement pattern of medical staff in each area; S22, activity frequency calculation submodule, used to pre-process the positioning trajectory, residence time and movement mode obtained by the personnel activity recognition unit, extract the number of activities per unit time, residence time and activity hot zone distribution, and calculate the regional activity frequency parameter : in, Indicates The length of time that medical staff stayed in the area Indicates Number of activities of medical staff, Indicates that the area is in the time interval The activation index of the active hotspot within is the preset weight coefficient, Indicates that in the observation time interval The number of medical personnel present in the area within the period; S23, patient status collection unit, used to collect information about whether the patient is in bed and whether the patient is in a resting state through the bed management system interface and the bed sensor, and obtain the bed occupancy rate to form the patient status parameter ,in , represents the number of patients in bed, Indicates the total number of beds in the region; S24, equipment operation status identification unit, the equipment operation status identification unit includes a data interface module with air purification equipment, medical gas equipment and negative pressure device, used to collect the switch status, operating power and alarm information of various types of equipment, and convert it into equipment state vector ; S25, a regional use level determination module, used to combine the , and Three groups of parameters were used to classify the regional functional status using hierarchical weighted scores; S26, a state vector generation module, used to perform data standardization and feature fusion according to the rules on the regional activity frequency parameters, patient state parameters and device state vectors to construct a regional state description vector .

3. The comprehensive air environment conditioning system for the hospital purification area according to claim 2 is characterized in that: The S3 specifically includes: S31, an input data receiving unit, used to receive sensing node data and a regional state description vector; S32, a data normalization processing submodule, used to normalize the sensing node data and the regional state description vector, convert different physical quantities into vector data representations within a unified dimension range, and generate a fusion input matrix; S33, an environmental map construction module, for constructing an undirected weighted graph structure based on the fusion input matrix and combining the physical space adjacency relationship, ventilation airflow path and functional relevance between the purification sub-areas; S34, graph convolution feature extraction submodule, for extracting features based on the undirected weighted graph structure and fusion input matrix , the state feature vector of each regional node is calculated through the graph convolutional network model : in, Represents and area nodes The set of adjacent nodes, For regional nodes With regional nodes The edge weights between Represents a regional node The corresponding fusion input matrix is is the weight parameter matrix in the graph convolutional network, is the activation function; S35, a key influencing factor screening unit, is used to identify the key influencing factors that affect the current regional air environment status based on the extracted state feature vector, using threshold discrimination and change trend analysis methods.

4. The comprehensive air environment conditioning system for the hospital purification area according to claim 3 is characterized in that: The S35 specifically includes: S351, feature change monitoring module, used to monitor the state feature vector Sliding window sampling is performed on the feature values ​​of each dimension to construct a historical sequence set ,in Indicates at the current moment No. eigenvalues, is the window length; S352, threshold determination unit, for each feature value Its historical average Calculate the deviation and obtain the change range index ,like Exceeding the preset sensitivity threshold , then mark the feature as a candidate influencing factor; S353, change trend scoring module, used to Calculate the slope of a linear fit for a time series trend in and the trend threshold set by the system Compare, if , the corresponding feature is regarded as a candidate factor with significant trend; S354, joint screening decision module, used to meet and Perform intersection operation on the characteristic indicators to generate a set of key influencing factors ,in, Indicates the first key influencing factors.

5. The comprehensive air environment conditioning system for the hospital purification area according to claim 4 is characterized in that: The S4 specifically includes: S41, fuzzy rule base construction unit, used to define fuzzy input variable set according to preset purification environment management strategy and the output variable set , establish the input-output fuzzy mapping rule set ; S42, fuzzy membership calculation module, used to set the key influencing factors Each eigenvalue in is mapped to a set of fuzzy input variables The membership function value of the input variable is calculated by using the triangular membership function. ; S43, inference engine module, used to calculate the value of the current fuzzy value matrix and the fuzzy mapping rule set , perform fuzzy logic reasoning and output fuzzy reasoning results ; S44, dynamic weight adjustment submodule, used to adjust the weight factor of each rule in the fuzzy mapping rule set based on historical control results and current area comfort feedback information , forming a weighted fuzzy rule set ,in Indicates Fuzzy rules, Indicates the update weight value corresponding to the rule, is the total number of rules in the fuzzy rule set; S45, defuzzification module, used to convert the fuzzy reasoning results Defuzzification is performed through the centroid method to generate a set of air conditioning parameters .

6. The comprehensive air environment conditioning system for a hospital purification area according to claim 1 is characterized in that: The S5 specifically includes: S61, physiological parameter acquisition unit, used to collect physiological parameters of medical staff and patients in the area through wearable and non-contact physiological sensor devices, the physiological parameters include skin temperature, heart rate, respiratory rate and skin conductivity, to form a physiological state vector ,in Indicates Real-time measurements of physiological parameters; S62, a subjective comfort rating input unit, used to collect the user's subjective comfort rating of the current air environment based on the mobile terminal, the panel device and the voice system, and the rating value corresponding to the subjective comfort rating is linearly normalized to form a subjective evaluation vector ,in Indicates The user's rating of the environment in the current cycle; S63, comfort feedback matrix generation module, used to convert the physiological state vector and subjective evaluation vector Perform weighted fusion to generate a comfort feedback matrix , the comfort feedback matrix represents the comprehensive comfort feedback degree of individual users to each environmental control variable: in, is the physiological state vector The normalized matrix expression is: is the subjective evaluation vector The normalized matrix expression is: is the fusion weight coefficient, which indicates the relative weight of physiological and subjective scores in the comprehensive evaluation. Indicates Class user Comfort feedback value of each air control parameter; S64, a strategy feedback optimization unit, used to convert the comfort feedback matrix The target error function is constructed by comparing the historical control output results of the fuzzy control system, and the weight factors of each rule in the fuzzy rule set are adjusted by the back propagation mechanism based on the target error function. Update and adjust to form a new set of weighted fuzzy rules ,in After optimization, The update weight factor of the rule, Indicates Fuzzy rules, is the total number of rules in the fuzzy rule set.

7. The comprehensive air environment conditioning system for the hospital purification area according to claim 6 is characterized in that: The S64 specifically includes: S641, historical output data cache module, used to periodically store air conditioning parameter sets ,in Indicates Air conditioning parameters in the current cycle The actual output value of S642, comfort target mapping module, used to convert the comfort feedback matrix The user feedback data in is converted into a set of target adjustment values ,in Indicates the user's expectation 2 air conditioning parameter setting values; S643, an error function construction unit, used to define a target error function based on the difference between the current cycle adjustment output value and the user expected value; S644, a weight updating module for updating a weight based on the target error function , adjust the weight factors of each rule in the fuzzy rule set according to the back-propagation update mechanism : in, represents the learning rate, For the current cycle The weight factor of the fuzzy rules, is the weight factor of each rule after update, Represents the gradient of the error with respect to the rule weight; S645, optimizing the rule set output unit, used to update the rule weights Applied to the next round of fuzzy control reasoning process to construct an updated weighted fuzzy rule set .

Images