Health-care sojourn environment air quality monitoring and purifying method and system based on Internet of Things

Abstract

The invention discloses a health-care sojourn environment air quality monitoring and purifying method and system based on the Internet of Things, and the method comprises the steps: dividing a dynamic microenvironment region, deploying a multi-mode sensing terminal, collecting indexes, and uploading the indexes to a cloud end; establishing a personalized dynamic standard according to the human body biological rhythm characteristics; generating an initial evaluation value through an environment-health coupling evaluation model, and judging whether the initial evaluation value deviates from a standard or not through a dynamic threshold deviation detection model; if so, predicting an air quality trend by using a space-time multi-dimensional trend deduction model, and generating a target evaluation value; and inputting the regional space purification model to generate a purification strategy, and linking the purification equipment to realize precise targeted purification. According to the method, the limitation of traditional fixed indexes is broken through, the upgrading from environmental monitoring to health response is realized, the health risk can be accurately evaluated, the trend can be predicted, the purification parameters are dynamically adjusted, a monitoring, analysis, prediction and purification closed loop is formed, the air quality regulation and control accuracy and energy efficiency are improved, and an efficient scheme is provided for healthy and nourishment of sojourn.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent air quality control based on the Internet of Things, and in particular to a method and system for monitoring and purifying the air quality of a health and wellness travel environment based on the Internet of Things. Background Art

[0002] With people's pursuit of healthy life, the air quality of the health and wellness travel environment is becoming more and more important; the health and wellness travel population is mostly elderly people, sub-healthy groups or rehabilitation patients, and their sensitivity to air quality is much higher than that of the general population. The traditional air quality monitoring and purification methods with fixed indicators can no longer meet the personalized and dynamic health and wellness needs; there is an urgent need for an intelligent monitoring and purification technology that can combine human health characteristics and dynamic environmental changes to achieve precise control of the air quality of the health and wellness travel environment and ensure the physical and mental health of travelers.

[0003] Traditional air quality monitoring and purification technologies typically employ fixed-point sensor placement and pre-set, uniform purification thresholds. While these technologies offer advantages in terms of simple system architecture and low deployment costs, they also lack the ability to account for individual differences among different populations, nor can they dynamically adjust to the functional needs of the environment and the patterns of human activity. This makes it difficult to accurately ensure air quality in wellness and travel settings, making traditional technologies particularly inadequate for specialized populations requiring differentiated health care.

[0004] Existing technologies attempt to adjust purification strategies based on environmental parameters, but most of them lack real-time monitoring and analysis of human physiological indicators, and have not established personalized air quality standards. At the same time, static area division is mostly used in spatial processing, which cannot adapt to the dynamic changes in functional areas caused by human activities in health and wellness travel environments. In addition, the generation of purification strategies lacks comprehensive consideration of the time and space dimensions, resulting in low purification efficiency or energy waste. Summary of the Invention

[0005] Based on the above content, this application discloses an intelligent monitoring and purification method for air quality in a health and wellness travel environment based on the Internet of Things, which solves the above-mentioned technical problems, including:

[0006] S1. Divide the wellness and living environment into dynamic microenvironmental zones based on spatial functions and human activity patterns. Deploy multimodal sensing terminals in each zone to obtain air quality and physiological indicators, which are then uploaded to the cloud management platform via the Internet of Things.

[0007] S2. Obtain the human biorhythm characteristics at different time periods and establish personalized dynamic health and wellness air quality standards through the cloud management platform;

[0008] S3. Obtain air and physiological indicators from the current cloud management platform and generate initial regional air-health assessment values ​​through the environment-health coupling assessment model;

[0009] S4. Inputting the obtained initial regional air-health assessment value and the personalized health-care air quality dynamic standard into a dynamic threshold deviation detection model to determine whether the initial regional air-health assessment value deviates from the personalized health-care air quality dynamic standard;

[0010] S5. If deviation occurs, obtain the initial regional air-health assessment value, combine the regional environmental conditions and human activity density, input the spatiotemporal multidimensional trend deduction model, predict the air quality trend in the future, and generate the target regional air-health assessment value;

[0011] S6. Obtain the air-health assessment value of the target area, input the pre-built regional space purification model, and generate the regional space purification strategy;

[0012] S7. After receiving the purification strategy, the purification equipment will execute linkage through the Internet of Things and dynamically adjust the direction, speed and purification range of the purification airflow according to the distribution and activity trajectory of people in the area to achieve precise targeted purification.

[0013] Preferably, the health and wellness travel environment is divided into dynamic micro-environment areas according to spatial functions and patterns of human activities in S1. Specifically, infrared thermal imaging sensors and millimeter-wave radar sensors deployed in the environment monitor the activity trajectories and aggregation density of people in real time, and the health and wellness travel environment is divided into basic living areas, rehabilitation and physiotherapy areas, meditation and rest areas, and social activity areas in combination with the functional attributes of the areas; at the same time, temporary functional areas are dynamically generated according to the real-time activity status of people, and the boundaries of each dynamic micro-environment area are defined by virtual electronic fence technology, and the range of the electronic fence is adjusted in real time according to changes in human activities.

[0014] Preferably, the human biorhythm characteristics of different time periods are obtained in S2 to establish a personalized dynamic standard for health and wellness air quality, specifically: obtaining the human biorhythm characteristic vector B and the correlation between regional air indicators, and establishing a personalized dynamic standard for health and wellness air quality S d , the formula is Where W is the biological rhythm feature weight matrix, λ k is the adjustment coefficient of the kth air index, C k is the standard reference value of the kth air index, m k is the number of air index types.

[0015] Preferably, the environment-health coupling assessment model in S3 is specifically as follows: a three-layer neural network structure is constructed including an environmental parameter layer, a physiological response layer, and a coupling assessment layer. The environmental parameter layer inputs the air index data X obtained by the current cloud management platform, and the physiological response layer inputs the physiological index data Y collected by the multimodal sensing terminal. The initial regional air-health assessment value EHI is calculated through the coupling assessment layer. The formula is: where α i is the weight of the i-th environmental indicator, β j is the weight of the jth physiological indicator, n is the number of environmental indicators, and m is the number of physiological indicators.

[0016] Preferably, the dynamic threshold deviation detection model in S4 is as follows: using LSTM to construct a time series feature extraction module, taking the initial regional air-health assessment value sequence and the personalized health care air quality dynamic standard sequence as input, learning the long-term dependency of the time series through the LSTM network, extracting the feature vector, and using the attention mechanism to construct a deviation calculation module, the formula is: Where W a 、W h 、W s are the corresponding weight matrices, b a is the bias vector, is the attention weight vector, and The characteristic vectors of the initial regional air-health assessment value sequence and the personalized health and wellness air quality dynamic standard sequence are calculated, and the deviation characteristic vector is calculated as follows: Build a dynamic threshold deviation detection model.

[0017] Preferably, the dynamic threshold deviation detection model is used to determine whether the initial regional air-health assessment value deviates from the personalized health care air quality dynamic standard, specifically: the deviation feature vector output by the dynamic threshold deviation detection model Use Mahalanobis distance to calculate the deviation from the preset standard vector The distance is: Where Σ is the covariance matrix of the deviation eigenvector; a deviation threshold τ is set. When MD>τ, the initial regional air-health assessment value is judged to deviate from the personalized health and wellness air quality dynamic standard, otherwise it is not deviated.

[0018] Preferably, the spatiotemporal multidimensional trend deduction model in S5 is constructed as follows: a model architecture is constructed through a graph convolutional network, and the regional initial air-health assessment value, regional real-time meteorological condition data, and personnel activity density data are used as node features. The spatial topological relationship of each dynamic microenvironment area is constructed as an adjacency matrix A, and a multi-layer graph convolution operation is performed. Extract spatial features, where is the adjacency matrix with self-loops added, is the degree matrix, X (l) is the node feature matrix of the lth layer, W (l) is the weight matrix of the lth layer, σ is the activation function, and the spatiotemporal multidimensional trend deduction model is obtained.

[0019] Preferably, the air quality trend in the future is predicted by the spatiotemporal multidimensional trend deduction model to generate the target area air-health assessment value, specifically: the current regional initial air-health assessment value, real-time meteorological condition data, and personnel activity density data are input into the spatiotemporal multidimensional trend deduction model, the spatial correlation features of each region are extracted, the time series features are learned through LSTM, and the spatiotemporal attention mechanism is used to calculate the weight of each feature in different spatiotemporal dimensions. Where G is the graph convolutional network output, L is the LSTM output, and W t 、W g 、W l is the corresponding weight matrix, b t As the bias vector, the air health assessment value HEV of the target area in the future Δt period is calculated and predicted t+Δt , the formula is:

[0020] Preferably, the air-health assessment value of the target area is obtained in S6, and the pre-built regional space purification model is input to generate the regional space purification strategy, specifically: the regional space purification model is constructed by using a double-depth Q network combined with a three-dimensional space, and the air-health assessment value of the target area, the regional space topology, and the purification equipment state parameters are used as the state space, and the operation of the purification equipment is used as the action space. The health and wellness travel environment is divided into a three-dimensional grid space through the three-dimensional space, and each grid corresponds to a state node. The formula is used Calculate the action value, where s is the current state, a is the current action, θ is the network parameter, r is the immediate reward γ is the discount factor, s ′ is the next state, a ′ is the next action, θ - The target network parameters are used to output the optimal regional space purification strategy through cumulative interactive learning.

[0021] The IoT-based intelligent air quality monitoring and purification system for health and wellness travel environments includes a dynamic microenvironment division module, a multimodal perception module, a cloud management module, an environment-health coupling assessment module, a dynamic threshold deviation detection module, a spatiotemporal multidimensional trend deduction module, a regional space purification model module, and a purification equipment linkage execution module. Each module achieves monitoring and purification through data interaction and control command transmission.

[0022] The dynamic microenvironment division module is connected to the multimodal perception module through an infrared thermal imaging sensor and a millimeter wave radar sensor, and is used to divide the environment into dynamic microenvironment areas according to spatial functions and human activity patterns and define virtual electronic fences;

[0023] The multimodal sensing module is connected to the cloud management module via the Internet of Things to collect air quality and physiological indicators in each area;

[0024] The cloud management module is respectively connected with the environment-health coupling assessment module, the dynamic threshold deviation detection module, the spatiotemporal multidimensional trend deduction module, and the regional space purification model module to establish personalized health and wellness air quality dynamic standards;

[0025] The environment-health coupling assessment module generates an initial regional air-health assessment value through a three-layer neural network structure; the dynamic threshold deviation detection module uses LSTM and attention mechanism to determine whether the assessment value deviates from the standard; the spatiotemporal multidimensional trend deduction module predicts air quality trends and generates target assessment values ​​through the fusion of graph convolutional network and LSTM; the regional space purification model module uses a dual deep Q network to generate purification strategies;

[0026] The purification equipment linkage execution module receives strategies through the Internet of Things and uses millimeter-wave radar and infrared thermal imaging data to dynamically adjust the direction, speed and range of the purification airflow.

[0027] Compared with the prior art, the technical solution of this application has the following technical effects:

[0028] By dividing the health and wellness travel environment into dynamic micro-environment areas and deploying multimodal sensing terminals, the present invention can obtain air indicators and human physiological indicators in real time. Combined with the personalized health and wellness air quality dynamic standards built on the cloud management platform, it realizes the upgrade from "environmental monitoring" to "health response". This method breaks through the limitations of traditional fixed indicator monitoring and can dynamically adjust the evaluation standards according to the differentiated needs of different health and wellness groups (such as patients with respiratory diseases and the elderly), so that air quality monitoring is more in line with individual health needs and provides travelers with accurate environmental health protection.

[0029] This paper utilizes a coupled environmental-health assessment model and a dynamic threshold deviation detection model, utilizing a three-layer neural network architecture and an LSTM-based attention mechanism to achieve a deep fusion analysis of air quality and physiological indicators. This mechanism accurately identifies the potential impacts of air quality changes on human health. Compared to traditional single environmental parameter assessments, it can detect health threats posed by abnormal air quality to specific populations earlier, providing a scientific basis for the timely initiation of purification measures and enhancing the system's ability to warn of health risks.

[0030] The combination of the spatiotemporal multidimensional trend deduction model of the present invention and the regional spatial purification model enables the system to have an intelligent closed-loop capability of "prediction-decision-making". By fusing and analyzing spatiotemporal data using a graph convolutional network and LSTM, the trend of air quality changes can be predicted in advance, and then a dual-depth Q network is used to generate a purification strategy, realizing the transition from "passive response" to "active prevention". This model can not only start the purification equipment in advance before pollution occurs, but also dynamically adjust the purification airflow according to the activity trajectory of personnel, thereby improving purification efficiency while reducing energy consumption, and optimizing the air quality control efficiency of the health and wellness travel environment.

[0031] The purification equipment linkage execution module of the present invention is designed to dynamically adjust purification parameters based on millimeter-wave radar and infrared thermal imaging data, realizing "precise targeted purification". Different from the traditional full-area unified purification method, this technology can adjust the direction, speed and range of the purified airflow in real time according to the distribution and activity trajectory of personnel, so that purification resources are concentrated in the personnel activity area. While ensuring the respiratory health of travelers, it avoids energy waste caused by ineffective purification, and provides an efficient and energy-saving air quality purification solution for health and wellness travel scenarios.

[0032] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application so that it can be implemented in accordance with the contents of the specification, and to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the following is a detailed description of the preferred embodiment of the present application in conjunction with the accompanying drawings.

[0033] Based on the detailed description of the specific embodiments of the present application in conjunction with the accompanying drawings below, those skilled in the art will become more aware of the above and other objects, advantages and features of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following will briefly introduce the drawings required for the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without inventive work. In all drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.

[0035] Figure 1 This is a flow chart of the air quality monitoring and purification method for the health and wellness travel environment based on the Internet of Things;

[0036] Figure 2 A flowchart for rapid silt removal in shallow waterways;

[0037] Figure 3This is a flow chart of the working principle of the dynamic threshold deviation detection model;

[0038] Figure 4 This is a flow chart for data processing and prediction of the spatiotemporal multidimensional trend deduction model;

[0039] Figure 5 Generate a flow chart for the regional spatial decontamination model strategy;

[0040] Figure 6 This is the structure diagram of the intelligent air quality monitoring and purification system for the health and wellness travel environment based on the Internet of Things;

[0041] Figure 7 This is a graph showing changes in PM2.5 concentration and respiratory rate during the activity period of people in the rehabilitation and physiotherapy area;

[0042] Figure 8 This is a comparison chart of the predicted and actual monitored PM2.5 concentration in the rehabilitation and physiotherapy area at 10:00 on the fifth day;

[0043] Figure 9 The following is a line graph comparing the energy consumption per unit area of ​​this system and the existing technology during the 30-day experimental period. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. In the following description, specific details such as specific configurations and components are provided only to help fully understand the embodiments of the present application. Therefore, it should be clear to those skilled in the art that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present application. In addition, for clarity and brevity, the description of known functions and structures has been omitted in the embodiments.

[0045] It should be understood that references throughout this specification to "one embodiment" or "this embodiment" mean that a particular feature, structure, or characteristic associated with the embodiment is included in at least one embodiment of the present application. Therefore, the appearance of "one embodiment" or "this embodiment" throughout this specification does not necessarily refer to the same embodiment. Furthermore, these particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0046] In addition, the present application may repeat reference numerals and / or letters in different examples. This repetition is for the purpose of simplicity and clarity and does not in itself indicate the relationship between the various embodiments and / or settings discussed.

[0047] The term "and / or" in this article is only a description of the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, B exists alone, and A and B exist at the same time. The term " / and" in this article describes another type of association object relationship, indicating that two relationships can exist. For example, A / and B can mean: A exists alone, and A and B exist alone. In addition, the character " / " in this article generally indicates that the previous and subsequent associated objects are in an "or" relationship.

[0048] The term "at least one" in this article is merely a description of the association relationship between associated objects, indicating that three relationships may exist. For example, at least one of A and B can mean: A exists alone, A and B exist at the same time, and B exists alone.

[0049] It should also be noted that, in this document, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "include," "comprises," or any other variations thereof are intended to cover non-exclusive inclusion.

[0050] Example 1

[0051] This embodiment mainly describes the air quality monitoring and purification method and system for health and wellness travel environment based on the Internet of Things. Figure 1 As shown, including:

[0052] S1. Divide the wellness and living environment into dynamic microenvironmental zones based on spatial functions and human activity patterns. Deploy multimodal sensing terminals in each zone to obtain air quality and physiological indicators, which are then uploaded to the cloud management platform via the Internet of Things.

[0053] S2. Obtain the human biorhythm characteristics at different time periods and establish personalized dynamic health and wellness air quality standards through the cloud management platform;

[0054] S3. Obtain air and physiological indicators from the current cloud management platform and generate initial regional air-health assessment values ​​through the environment-health coupling assessment model;

[0055] S4. Inputting the obtained initial regional air-health assessment value and the personalized health-care air quality dynamic standard into a dynamic threshold deviation detection model to determine whether the initial regional air-health assessment value deviates from the personalized health-care air quality dynamic standard;

[0056] S5. If deviation occurs, obtain the initial regional air-health assessment value, combine the regional environmental conditions and human activity density, input the spatiotemporal multidimensional trend deduction model, predict the air quality trend in the future, and generate the target regional air-health assessment value;

[0057] S6. Obtain the air-health assessment value of the target area, input the pre-built regional space purification model, and generate the regional space purification strategy;

[0058] S7. After receiving the purification strategy, the purification equipment will execute linkage through the Internet of Things and dynamically adjust the direction, speed and purification range of the purification airflow according to the distribution and activity trajectory of people in the area to achieve precise targeted purification.

[0059] Furthermore, S1 divides the health and wellness travel environment into dynamic micro-environment areas according to spatial functions and patterns of human activities. Specifically, infrared thermal imaging sensors and millimeter-wave radar sensors deployed in the environment monitor the activity trajectories and aggregation density of people in real time. Combined with the functional attributes of the area, the health and wellness travel environment is divided into basic living areas, rehabilitation and physiotherapy areas, meditation and rest areas, and social activity areas. At the same time, temporary functional areas are dynamically generated according to the real-time activity status of people. The boundaries of each dynamic micro-environment area are defined by virtual electronic fence technology, and the range of the electronic fence is adjusted in real time according to changes in human activities.

[0060] Furthermore, in S2, the human biorhythm characteristics of different time periods are obtained to establish a personalized dynamic standard for health and wellness air quality. Specifically, the correlation between the human biorhythm characteristic vector B and the regional air index is obtained to establish a personalized dynamic standard for health and wellness air quality S. d , the formula is Where W is the biological rhythm feature weight matrix, λ k is the adjustment coefficient of the kth air index, C k is the standard reference value of the kth air index, m k is the number of air index types.

[0061] Further, if Figure 2 As shown in Figure 3, the environment-health coupling assessment model in S3 is specifically as follows: a three-layer neural network structure is constructed, which includes an environmental parameter layer, a physiological response layer, and a coupling assessment layer. The environmental parameter layer inputs the air index data X obtained by the current cloud management platform, and the physiological response layer inputs the physiological index data Y collected by the multimodal sensing terminal. The coupling assessment layer calculates the initial regional air-health assessment value EHI, and the formula is: where α i is the weight of the i-th environmental indicator, β j is the weight of the jth physiological indicator, n is the number of environmental indicators, and m is the number of physiological indicators.

[0062] Further, if Figure 3 As shown in Figure 4, the dynamic threshold deviation detection model in S4 is as follows: LSTM is used to build a time series feature extraction module, which takes the initial regional air-health assessment value sequence and the personalized health and wellness air quality dynamic standard sequence as input. The LSTM network learns the long-term dependency of the time series, extracts the feature vector, and uses the attention mechanism to build a deviation calculation module. The formula is: Where W a 、W h 、W s are the corresponding weight matrices, b a is the bias vector, is the attention weight vector, and The characteristic vectors of the initial regional air-health assessment value sequence and the personalized health and wellness air quality dynamic standard sequence are calculated, and the deviation characteristic vector is calculated as follows: Build a dynamic threshold deviation detection model.

[0063] Furthermore, the dynamic threshold deviation detection model is used to determine whether the initial regional air-health assessment value deviates from the personalized health and wellness air quality dynamic standard. Specifically, the deviation feature vector output by the dynamic threshold deviation detection model is Use Mahalanobis distance to calculate the deviation from the preset standard vector The distance is: Where Σ is the covariance matrix of the deviation eigenvector; a deviation threshold τ is set. When MD>τ, the initial regional air-health assessment value is judged to deviate from the personalized health and wellness air quality dynamic standard, otherwise it is not deviated.

[0064] Further, if Figure 4 As shown in the figure, the spatiotemporal multidimensional trend deduction model in S5 is constructed as follows: the model architecture is constructed through the graph convolution network, the regional initial air-health assessment value, the regional real-time meteorological condition data, and the personnel activity density data are used as node features, and the spatial topological relationship of each dynamic microenvironment area is constructed as an adjacency matrix A. Extract spatial features, where is the adjacency matrix with self-loops added, is the degree matrix, X (l) is the node feature matrix of the lth layer, W (l) is the weight matrix of the lth layer, σ is the activation function, and the spatiotemporal multidimensional trend deduction model is obtained.

[0065] Further, if Figure 5As shown in the figure, the air quality trend in the future is predicted by the spatiotemporal multidimensional trend deduction model to generate the air-health assessment value of the target area. Specifically, the initial air-health assessment value of the region at the current moment, real-time meteorological condition data, and personnel activity density data are input into the spatiotemporal multidimensional trend deduction model, the spatial correlation features of each region are extracted, the time series features are learned through LSTM, and the spatiotemporal attention mechanism is used to calculate the weight of each feature in different spatiotemporal dimensions. Where G is the graph convolutional network output, L is the LSTM output, and W t 、W g 、W l is the corresponding weight matrix, b t As the bias vector, the air health assessment value HEV of the target area in the future Δt period is calculated and predicted t+Δt , the formula is:

[0066] Furthermore, in S6, the air-health assessment value of the target area is obtained and the pre-built regional space purification model is input to generate the regional space purification strategy. Specifically, the regional space purification model is constructed by combining a double-depth Q network with a three-dimensional space. The air-health assessment value of the target area, the regional space topology, and the state parameters of the purification equipment are used as the state space, and the operation of the purification equipment is used as the action space. The health and wellness travel environment is divided into a three-dimensional grid space through the three-dimensional space. Each grid corresponds to a state node. The formula is used. Calculate the action value, where s is the current state, a is the current action, θ is the network parameter, r is the immediate reward γ is the discount factor, s ′ is the next state, a ′ is the next action, θ - The target network parameters are used to output the optimal regional space purification strategy through cumulative interactive learning.

[0067] This embodiment describes in detail how this technical solution achieves accurate characterization of the health and wellness travel environment through dynamic microenvironment division combined with multimodal perception; introduces human biorhythm characteristics to construct personalized dynamic standards, breaking through the limitations of traditional unified thresholds; and utilizes deep learning models to achieve intelligent decision-making throughout the entire process from evaluation and prediction to purification. In particular, dynamic purification strategies are generated through spatiotemporal multidimensional analysis and reinforcement learning, solving the problems of static processing and poor individual adaptability in existing technologies, and significantly improving the accuracy and intelligence of air quality control in the health and wellness travel environment.

[0068] Example 2: This example describes in detail an intelligent monitoring and purification system for air quality in a health and wellness travel environment based on the Internet of Things. Figure 6 As shown, specifically:

[0069] The IoT-based intelligent air quality monitoring and purification system for health and wellness travel environments includes a dynamic microenvironment division module, a multimodal perception module, a cloud management module, an environment-health coupling assessment module, a dynamic threshold deviation detection module, a spatiotemporal multidimensional trend deduction module, a regional space purification model module, and a purification equipment linkage execution module. Each module achieves monitoring and purification through data interaction and control command transmission.

[0070] The dynamic microenvironment division module is connected to the multimodal perception module through an infrared thermal imaging sensor and a millimeter wave radar sensor, and is used to divide the environment into dynamic microenvironment areas according to spatial functions and human activity patterns and define virtual electronic fences;

[0071] The multimodal sensing module is connected to the cloud management module via the Internet of Things to collect air quality and physiological indicators in each area;

[0072] The cloud management module is respectively connected with the environment-health coupling assessment module, the dynamic threshold deviation detection module, the spatiotemporal multidimensional trend deduction module, and the regional space purification model module to establish personalized health and wellness air quality dynamic standards;

[0073] The environment-health coupling assessment module generates an initial regional air-health assessment value through a three-layer neural network structure; the dynamic threshold deviation detection module uses LSTM and attention mechanism to determine whether the assessment value deviates from the standard; the spatiotemporal multidimensional trend deduction module predicts air quality trends and generates target assessment values ​​through the fusion of graph convolutional network and LSTM; the regional space purification model module uses a dual deep Q network to generate purification strategies;

[0074] The purification equipment linkage execution module receives strategies through the Internet of Things and uses millimeter wave radar and infrared thermal imaging data to dynamically adjust the direction, speed and range of the purification airflow;

[0075] Furthermore, the dynamic microenvironment division module uses infrared thermal imaging and millimeter-wave radar sensors to capture the activity trajectories of people and the functional attributes of the area in real time, divide the health and wellness travel environment into dynamic functional areas such as basic living areas and rehabilitation and physiotherapy areas, and define the boundaries with virtual electronic fences; the multimodal perception module deploys sensor terminals that integrate air indicators (PM2.5, VOC, etc.) and human physiological indicators (heart rate, respiratory rate) in each area, and uploads multi-dimensional data to the cloud management module through the Internet of Things, forming a basic link for "environment-health" data interaction.

[0076] Furthermore, as the core of the system, the cloud management module assumes the central functions of data processing and strategy decision-making. Based on multimodal perception data, combined with information such as personnel age and health records, it constructs personalized dynamic standards for health and wellness air quality through biorhythm feature extraction and grey correlation analysis, so that different groups of people (such as the elderly and patients with respiratory diseases) have their own exclusive air quality threshold system; at the same time, the cloud management module is deeply connected with the environment-health coupling assessment module, dynamic threshold deviation detection module, etc., and generates air-health assessment values ​​through a three-layer neural network. Then, the LSTM and attention mechanism algorithm is used to judge the degree of deviation between the assessment value and the dynamic standard, realizing intelligent analysis from data collection to health risk assessment.

[0077] Furthermore, the spatiotemporal multi-dimensional trend deduction module and the regional spatial purification model module constitute the system's "intelligent decision-making engine." The former extracts the spatial topological features of each dynamic microenvironment through a graph convolutional network, combines LSTM to capture time series patterns, and integrates the spatiotemporal attention mechanism to predict future air quality trends and generate air-health assessment values ​​for the target area. The latter adopts a dual-depth Q-network reinforcement learning framework, taking the target assessment value, spatial topological structure, and purification equipment state parameters as inputs, and outputs the optimal purification strategy through three-dimensional grid space modeling and action value calculation, realizing a technological upgrade from "passive response to pollution" to "active prediction and control," ensuring the accuracy and foresight of the purification strategy.

[0078] Furthermore, the purification equipment linkage execution module serves as the physical execution terminal of the system. It receives purification strategies generated in the cloud through the Internet of Things, and dynamically adjusts the direction, speed, and coverage of the purified airflow based on the real-time feedback of personnel distribution data from millimeter-wave radar and infrared thermal imaging. For example, when the system detects that people are gathering in a meditation and rest area, it automatically designates that area as a key purification zone, adjusts the airflow direction of the fresh air system and the negative oxygen ion generator, forms a targeted purification zone, and avoids energy waste from full-area purification. This "personnel activity-equipment response" linkage mechanism not only ensures the respiratory health of the health care population, but also achieves energy-saving optimization through intelligent control, which meets the refined needs of health care travel scenarios.

[0079] This detailed implementation description uses dynamic microenvironment division and multimodal perception to achieve real-time monitoring of environmental and human physiological indicators. Combined with cloud-based personalized standard construction, multi-level intelligent analysis models and reinforcement learning strategy generation, it can accurately assess health risks and predict air quality trends. It can also link purification equipment to dynamically adjust purification parameters according to personnel activities, realizing a full-process intelligent closed loop from monitoring, analysis, prediction to precise purification, thereby improving the accuracy and energy efficiency of air quality control in health and wellness travel environments.

[0080] Based on Examples 1 and 2, this example describes in detail the specific implementation process and comparative effects of the monitoring and purification of the present application, specifically:

[0081] A three-story residential building in a wellness and wellness community, with a total floor area of ​​approximately 2,800 square meters, was selected as the experimental site. The first floor housed public areas (including a lobby, dining room, and social areas), the second floor served as a rehabilitation and physiotherapy area (including a recovery room and a meditation area), and the third floor housed accommodations (including 20 bedrooms and a recreational area). The experiment lasted 30 days and covered various weather conditions and population density.

[0082] Experimental equipment deployment and dynamic micro-environment division were carried out. 12 infrared thermal imaging sensors and 8 millimeter-wave radar sensors were deployed to monitor the activity trajectory and gathering density of people. Multimodal perception module: 26 multi-parameter air quality monitoring terminals were deployed in each area (each integrated with a PM2.5 sensor (detection range 0-1000μg / m 3 , accuracy ±5μg / m 3 ), PM10 sensor (detection range 0-2000μg / m 3 , accuracy ±10μg / m 3 ), VOC sensor (detection range 0-20ppm, accuracy ±0.1ppm), carbon dioxide sensor (detection range 0-5000ppm, accuracy ±30ppm), negative oxygen ion sensor (detection range 0-50000 / cm 3 , accuracy ±10%) and human biological signal sensor (collecting physiological indicators such as heart rate and respiratory rate, sampling frequency 10Hz);

[0083] Purification equipment linkage execution module: deploy 32 different types of purification equipment, including high-efficiency air purifiers (CADR value 350-500m 3 / h) 18 units, fresh air system (air volume 200-800m 3 / h) 8 sets, negative oxygen ion generator (release amount 1-5×10 6 pieces / cm 3 )6 units.

[0084] Select comparative technical equipment:

[0085] Comparison Technology 1: Traditional fixed-point monitoring + unified threshold purification system, deploying 8 conventional air quality sensors (including PM2.5, PM10, CO2 sensors), 12 purification equipment (air purifiers, CADR value 200-300m 3 / h, using a fixed threshold (PM2.5 ≥ 35 μg / m 3 , start purification when CO2≥1000ppm).

[0086] Comparison Technology 2: IoT-based static area monitoring and purification system. This system deploys 15 IoT sensors (including PM2.5, VOC, and CO2 sensors), divides the air into fixed areas (such as bedrooms and living rooms), and uses preset purification strategies with no dynamic adjustment mechanism.

[0087] Comparison Technology 3: Single-Parameter Optimization Purification System. This system focuses on monitoring PM2.5, deploying 10 PM2.5 sensors. The purification equipment only regulates PM2.5, ignoring other air quality indicators and human physiological data.

[0088] The experiment was divided into an experimental group and a control group. The experimental group used this application to monitor and purify air quality. The control group used comparison technology 1, comparison technology 2, and comparison technology 3 to monitor and purify under the same conditions.

[0089] Baseline data was collected at the experimental site, including initial air quality and personnel activity patterns. Within 30 days, different health and wellness travel scenarios (such as daily activities, rehabilitation therapy, sleep, etc.) were simulated daily, and the monitoring data of each system, the implementation of purification strategies, and energy consumption data were recorded. The massive amount of collected data was collated and analyzed, and the experimental results of each group were compared.

[0090] During the experiment, this application used infrared thermal imaging and millimeter-wave radar sensors to monitor personnel activities in real time, dynamically dividing a total of 136 regular areas, including basic living areas, rehabilitation and physiotherapy areas, meditation and rest areas, social activity areas, as well as temporary functional areas such as temporary yoga areas and reading corners. Taking the 15th day as an example, the specific data is shown in Table 1:

[0091] Table 1 Dynamic microenvironmental division and personnel activity data of the health and wellness travel environment on the 15th day

[0092]

[0093] like Figure 7 As shown in the figure, it can be seen that the amount of air index and physiological index data collected by the multimodal sensing module every day is huge. Taking PM2.5 concentration as an example, 26 monitoring points in the entire area collected a total of 26×24×60×30=1123200 valid data in 30 days. The data distribution is shown in the figure. It can be seen from the figure that during the intensive activity period of 10:00-11:00 in the rehabilitation and physiotherapy area, the PM2.5 concentration fluctuated slightly (from 18.6μg / m 3 Increased to 22.3 μg / m 3 ), the system captured this change in time through multimodal perception, and combined with physiological indicators such as the respiratory rate of the personnel collected during the same period (the average increased from 16.2 times / minute to 18.7 times / minute), it provided multi-dimensional data support for subsequent health assessments.

[0094] This application collects 30 consecutive days of physiological indicators (heart rate, respiratory rate, body temperature, etc.) and sleep cycle data from 20 travelers of different age groups (55-75 years old) and different health conditions, uses Fourier transform to extract biorhythm characteristics, and establishes personalized dynamic standards for health and wellness air quality. Taking one of the travelers (numbered P05) who is 65 years old and has mild respiratory disease as an example, the comparison of the personalized dynamic standards for some of his air indicators with the traditional fixed standards is shown in Table 2 below, specifically:

[0095] Table 2 Comparison between personalized dynamic air quality standards and traditional fixed standards for travelers No. P05

[0096]

[0097] At the same time, the environment-health coupled assessment module uses a three-layer neural network structure to integrate and analyze air and physiological indicators to generate an initial regional air-health assessment value (EHI). During the 30-day experiment, the system generated a total of 26 × 24 × 60 × 30 = 1,123,200 EHI values, of which 1,386 deviations from the personalized dynamic standard were detected. Taking the social activity area from 2:00 PM to 4:00 PM on the 20th day as an example, the specific data is shown in Table 3:

[0098] Table 3 Deviations between air quality, physiological indicators, and EHI values ​​in the social activity area from 14:00 to 16:00 on the 20th day

[0099]

[0100] The dynamic threshold deviation detection module uses LSTM and the attention mechanism to calculate the deviation feature vector and judges the degree of deviation through the Mahalanobis distance. Calculation shows that the Mahalanobis distance MD at 15:00 is 2.87, which exceeds the preset threshold τ = 2.5. The system determines it as a significant deviation, triggering subsequent trend prediction and purification strategy generation.

[0101] The spatiotemporal multidimensional trend deduction module predicts future air quality trends by integrating a graph convolutional network with an LSTM. Taking the PM2.5 concentration in the rehabilitation and physiotherapy area predicted for the next hour at 10:00 AM on the fifth day as an example, a comparison between the predicted data and the actual monitored data is shown in Table 4. Specifically:

[0102] Table 4 Comparison of PM2.5 concentration prediction and actual monitoring data in the rehabilitation therapy area for the next hour at 10:00 on the 5th day

[0103]

[0104]

[0105] like Figure 8As shown in the figure, the prediction of the PM2.5 concentration in the rehabilitation therapy area for the next hour at 10:00 on the 5th day is compared with the actual monitoring. The blue curve is the predicted value and error range, the red square is the actual monitoring data, and the green bar graph is the error percentage. The prediction is consistent with the actual trend, and the error is between 2.98% and 4.29%, which reflects the accuracy of the model.

[0106] The regional space purification model module uses a dual-depth Q network to generate purification strategies. During the 30-day experiment, the optimal purification strategy was generated 4768 times. Taking the purification strategy of the restaurant during lunch time on the 18th day as an example, the system generated the optimal purification strategy based on the crowd density (0.58 people / square meter), air quality index (PM2.5 = 24.7μg / m 3 , CO2=915ppm) and the predicted trend, the following strategy is generated:

[0107] Start two high-efficiency air purifiers (CADR values ​​are 420m 3 / h and 380m 3 / h), the wind speed is set to medium (the wind volume is 280m 3 / h and 250m 3 / h);

[0108] Open the fresh air system (air volume 600m 3 / h), the airflow direction is adjusted from the restaurant entrance to the window;

[0109] Start the negative oxygen ion generator (release amount 3×10 6 pieces / cm 3 ), focusing on areas where people gather;

[0110] The purification equipment linkage execution module dynamically adjusts the purification airflow based on the purification strategy, combined with millimeter-wave radar and infrared thermal imaging data. Taking the meditation and rest area on the afternoon of the seventh day as an example, when the system detected five people meditating in this area, the purification parameters were adjusted in real time as shown in Table 5:

[0111] Table 5: Parameter adjustment table for purification equipment in the meditation and rest area on the afternoon of the 7th day

[0112]

[0113] After the above adjustments, the air quality in the area was significantly improved within 15 minutes. The specific data are shown in Table 6:

[0114] Table 6 Changes in air quality indicators before and after adjustment of purification equipment in the meditation and quiet area

[0115] index Before adjustment 5 minutes after adjustment 10 minutes after adjustment 15 minutes after adjustment <![CDATA[PM2.5(μg / m 3 )]]> 19.2 15.7 12.3 9.8 VOC (ppm) 0.35 0.30 0.25 0.22 <![CDATA[CO2(ppm)]]> 780 720 680 650 <![CDATA[Negative oxygen ions (per cm 3 )]]> 600 1200 1800 2200

[0116] At the same time, the improvement effect of the comparative technology in the 30-day experiment was obtained, and the improvement effect of the main air indicators of the present application and the comparative technology were compared, as shown in Table 7:

[0117] Table 7 Comparison of the improvement effect of main air indicators between this system and existing technology

[0118] Air Index This system Prior Art 1 Prior Art 2 Prior Art 3 <![CDATA[PM2.5 average concentration (μg / m 3 )]]> 12.6±3.2 21.5±4.8 18.7±4.1 15.8±3.9 <![CDATA[PM2.5 compliance rate (≤35 μg / m 3 )]]> 98.6％ 85.2％ 91.3％ 94.7％ Average VOC concentration (ppm) 0.21±0.05 0.38±0.09 0.32±0.07 0.28±0.06 VOC compliance rate (≤0.6ppm) 100％ 92.4％ 96.7％ 98.5％ <![CDATA[Average CO2 concentration (ppm)]]> 680±120 890±150 820±130 750±140 <![CDATA[CO2 compliance rate (≤1000 ppm)]]> 100％ 95.8％ 98.2％ 99.3％ <![CDATA[Average concentration of negative oxygen ions (number / cm 3 )]]> 1560±420 650±180 820±210 780±190

[0119] The purification efficiency of this application and the comparison technology (PM2.5 from 50μg / m 3 Reduced to 35 μg / m 3 The comparison of the required time (taking the time required for purification as an example) and the response time (the time from the indicator exceeding the standard to the start of purification) is shown in Table 8:

[0120] Table 8 Comparison of purification efficiency and response time between this system and existing technology

[0121] Technology Type Purification efficiency (minutes) Response time (seconds) This system 8.7±1.2 12.5±3.8 Prior Art 1 15.6±2.3 45.2±10.5 Prior Art 2 12.3±1.8 28.7±8.2 Prior Art 3 10.9±1.5 32.4±9.1

[0122] During the 30-day experiment, the energy consumption of each system (taking the power consumption per square meter of building area as an example) was compared, as shown in Table 9 below:

[0123] Table 9 Comparison of energy consumption of this system and existing technology during 30-day experiment

[0124] Technology Type Total energy consumption (kWh) <![CDATA[Energy consumption per unit area (kWh / m 2 )]]> This system 1856.3 0.663 Prior Art 1 2532.7 0.904 Prior Art 2 2215.4 0.791 Prior Art 3 2087.6 0.746

[0125] from Figure 9 As shown in the figure, it can be clearly seen that the energy consumption curve of this application is always at the lowest level during the 30-day experimental period, and the energy consumption per unit area is only 0.663kWh / m 2 , a 26.7% reduction compared to Comparative Technology 1, a 16.2% reduction compared to Comparative Technology 2, and a 11.1% reduction compared to Comparative Technology 3. Specifically, during daily energy consumption fluctuations, this application can adjust the operating status of the purification equipment in real time according to the density of personnel activities and air quality, avoiding ineffective energy consumption. For example, during the night when people are resting, this application will divide the bedroom area into dynamic microenvironments and only start targeted purification in the area near the bed where the person is located, reducing energy consumption by about 40% compared to Comparative Technology 1.

[0126] Further analysis of the energy consumption distribution in different functional areas shows that this application reduces energy consumption by 18.3%-25.6% compared with comparison technology 2 in areas with high air quality requirements such as rehabilitation and physiotherapy areas and meditation and rest areas, while ensuring that the air quality meets the standards. Comparison technology 1 uses a fixed threshold for full-area purification, maintaining high-power operation even in sparsely populated periods, resulting in significantly higher energy consumption than this application. Although comparison technology 3 is only optimized for PM2.5, it ignores the purification needs of other indicators such as VOC and CO2. In actual operation, in order to maintain comprehensive air quality, the equipment is frequently started and stopped, which increases energy consumption fluctuations and total energy consumption.

[0127] From the perspective of energy consumption composition, the purification equipment linkage execution module of this application reduces the average operating power of high-efficiency air purifiers by 22.4% by dynamically adjusting the direction, speed and range of the airflow, and the air volume adjustment accuracy of the fresh air system is improved to ±50m3 / h, which saves 31.7% energy compared to the fixed air volume mode of comparative technology 1. At the same time, under the intelligent regulation of this application, the negative oxygen ion generator only operates at an appropriate intensity during the period of personnel activity, and the energy consumption is reduced by 43.2% compared to the continuous operation mode of prior art 2. These data fully demonstrate that this application has achieved the optimal balance between purification efficiency and energy consumption control through dynamic microenvironment division, multimodal perception and intelligent strategy generation, and has significant energy-saving advantages in health and wellness travel scenarios.

[0128] This embodiment describes in detail the significant technical effects of this system in the experiment. Dynamic microenvironment segmentation and multimodal sensing make the average PM2.5 concentration reach 12.6±3.2μg / m 3 The spatiotemporal prediction model shortened the purification response time to 12.5±3.8 seconds, and within 15 minutes, PM2.5 was reduced from 19.2μg / m 3 Reduced to 9.8 μg / m 3 The equipment linkage realizes precise targeted purification, with energy consumption per unit area of ​​0.663kWh / m 2 , which is 26.7% lower than the existing technology 1. The system improves the accuracy and energy efficiency of air quality control through the "monitoring-assessment-prediction-purification" closed loop, providing an efficient solution for health and wellness travel.

[0129] The above are only preferred embodiments of the present invention, which do not limit the scope of protection of the present invention. For those skilled in the art, the present invention can be modified and varied in various ways. Any changes, modifications, replacements, integrations and parameter changes to these embodiments through conventional substitutions or that can achieve the same functions without departing from the principles and spirit of the present invention fall within the scope of protection of the present invention.

Claims

1. An intelligent monitoring and purification method for air quality in a health and wellness travel environment based on the Internet of Things, characterized in that: include: S1. Divide the wellness and living environment into dynamic microenvironmental zones based on spatial functions and human activity patterns. Deploy multimodal sensing terminals in each zone to obtain air quality and physiological indicators, which are then uploaded to the cloud management platform via the Internet of Things. S2. Obtain the human biorhythm characteristics at different time periods and establish personalized dynamic health and wellness air quality standards through the cloud management platform; S3. Obtain air and physiological indicators from the current cloud management platform and generate initial regional air-health assessment values ​​through the environment-health coupling assessment model; S4. Inputting the obtained initial regional air-health assessment value and the personalized health-care air quality dynamic standard into a dynamic threshold deviation detection model to determine whether the initial regional air-health assessment value deviates from the personalized health-care air quality dynamic standard; S5. If deviation occurs, obtain the initial regional air-health assessment value, combine the regional environmental conditions and human activity density, input the spatiotemporal multidimensional trend deduction model, predict the air quality trend in the future, and generate the target regional air-health assessment value; S6. Obtain the air-health assessment value of the target area, input the pre-built regional space purification model, and generate the regional space purification strategy; S7. After receiving the purification strategy, the purification equipment will execute linkage through the Internet of Things and dynamically adjust the direction, speed and purification range of the purification airflow according to the distribution and activity trajectory of people in the area to achieve precise targeted purification.

2. The method for intelligent monitoring and purification of air quality in a health and wellness travel environment based on the Internet of Things according to claim 1 is characterized in that: In the said S1, the health care and living environment is divided into dynamic micro-environment areas according to the spatial functions and the patterns of human activities. Specifically, the infrared thermal imaging sensors and millimeter wave radar sensors deployed in the environment are used to monitor the activity trajectories and aggregation density of people in real time. Combined with the functional attributes of the area, the health care and living environment is divided into basic living area, rehabilitation and physiotherapy area, meditation and rest area, and social activity area. At the same time, temporary functional areas are dynamically generated according to the real-time activity status of people. The boundaries of each dynamic micro-environment area are defined by virtual electronic fence technology, and the range of the electronic fence is adjusted in real time according to changes in human activities.

3. The method for intelligent monitoring and purification of air quality in a health and wellness travel environment based on the Internet of Things according to claim 1 is characterized in that: In said S2, the human biorhythm characteristics of different time periods are obtained to establish a personalized dynamic standard for health-care air quality. Specifically, the human biorhythm characteristic vector B is obtained and the correlation between the regional air index is established to establish a personalized dynamic standard for health-care air quality S. d , the formula is Where W is the biological rhythm feature weight matrix, λ k is the adjustment coefficient of the kth air index, C k is the standard reference value of the kth air index, m k is the number of air index types.

4. The method for intelligent monitoring and purification of air quality in a health and wellness travel environment based on the Internet of Things according to claim 1 is characterized in that: The environment-health coupling assessment model in S3 is specifically constructed as follows: a three-layer neural network structure is constructed, which includes an environmental parameter layer, a physiological response layer, and a coupling assessment layer. The environmental parameter layer inputs the air index data X obtained by the current cloud management platform, and the physiological response layer inputs the physiological index data Y collected by the multimodal sensing terminal. The coupling assessment layer calculates the initial regional air-health assessment value EHI. The formula is: where α i is the weight of the i-th environmental indicator, β j is the weight of the jth physiological indicator, n is the number of environmental indicators, and m is the number of physiological indicators.

5. The method for intelligent monitoring and purification of air quality in a health and wellness travel environment based on the Internet of Things according to claim 1 is characterized in that: The dynamic threshold deviation detection model in S4 is as follows: LSTM is used to construct a time series feature extraction module, which takes the initial regional air-health assessment value sequence and the personalized health and wellness air quality dynamic standard sequence as input, learns the long-term dependency of the time series through the LSTM network, extracts the feature vector, and uses the attention mechanism to construct a deviation calculation module. The formula is: Where W a 、W h 、W s are the corresponding weight matrices, b a is the bias vector, is the attention weight vector, and The characteristic vectors of the initial regional air-health assessment value sequence and the personalized health and wellness air quality dynamic standard sequence are calculated, and the deviation characteristic vector is calculated as follows: Build a dynamic threshold deviation detection model.

6. The method for intelligent monitoring and purification of air quality in a health and wellness travel environment based on the Internet of Things according to claim 1 or 5, characterized in that: The dynamic threshold deviation detection model is used to determine whether the initial regional air-health assessment value deviates from the personalized health and wellness air quality dynamic standard. Specifically, the deviation feature vector output by the dynamic threshold deviation detection model is Use Mahalanobis distance to calculate the deviation from the preset standard vector The distance is: Where Σ is the covariance matrix of the deviation eigenvector; a deviation threshold τ is set. When MD>τ, the initial regional air-health assessment value is judged to deviate from the personalized health and wellness air quality dynamic standard, otherwise it is not deviated.

7. The method for intelligent monitoring and purification of air quality in a health and wellness travel environment based on the Internet of Things according to claim 1 is characterized in that: The spatiotemporal multidimensional trend deduction model in S5 is constructed as follows: a model architecture is constructed through a graph convolutional network, and the regional initial air-health assessment value, regional real-time meteorological condition data, and personnel activity density data are used as node features. The spatial topological relationship of each dynamic microenvironment area is constructed as an adjacency matrix A, and a multi-layer graph convolution operation is performed. Extract spatial features, where is the adjacency matrix with self-loops added, I is the identity matrix, is the degree matrix, X (l) is the node feature matrix of the lth layer, W (l) is the weight matrix of the lth layer, σ is the activation function, and the spatiotemporal multidimensional trend deduction model is obtained.

8. The method for intelligent monitoring and purification of air quality in a health and wellness travel environment based on the Internet of Things according to claim 1 or 7, characterized in that: The spatiotemporal multidimensional trend deduction model is used to predict future air quality trends and generate air-health assessment values ​​for target areas. Specifically, the current regional initial air-health assessment value, real-time meteorological conditions data, and human activity density data are input into the spatiotemporal multidimensional trend deduction model. The spatial correlation features of each region are extracted, the time series features are learned through LSTM, and the spatiotemporal attention mechanism is used to calculate the weights of each feature in different spatiotemporal dimensions. Where G is the graph convolutional network output, L is the LSTM output, and W t 、W g 、W l is the corresponding weight matrix, b t As the bias vector, the air health assessment value HEV of the target area in the future Δt period is calculated and predicted t+Δt , the formula is:

9. The method for intelligent monitoring and purification of air quality in a health and wellness travel environment based on the Internet of Things according to claim 1 is characterized in that: In the step S6, the target area air-health assessment value is obtained, and the pre-built regional space purification model is input to generate the regional space purification strategy. Specifically, the regional space purification model is constructed by combining a double-depth Q network with a three-dimensional space. The target area air-health assessment value, the regional space topology, and the purification equipment state parameters are used as the state space. The operation of the purification equipment is used as the action space. The health and wellness travel environment is divided into a three-dimensional grid space through the three-dimensional space. Each grid corresponds to a state node. The formula is used. Calculate the action value, where s is the current state, α is the current action, θ is the network parameter, r is the immediate reward γ is the discount factor, s ′ is the next state, a ′ is the next action, θ - The target network parameters are used to output the optimal regional space purification strategy through cumulative interactive learning.

10. An intelligent air quality monitoring and purification system for a health and wellness travel environment based on the Internet of Things, applicable to any one of claims 1-9, characterized in that: It includes a dynamic microenvironment division module, a multimodal perception module, a cloud management module, an environment-health coupling assessment module, a dynamic threshold deviation detection module, a spatiotemporal multidimensional trend deduction module, a regional space purification model module, and a purification equipment linkage execution module. Each module realizes monitoring and purification through data interaction and control command transmission; The dynamic microenvironment division module is connected to the multimodal perception module through an infrared thermal imaging sensor and a millimeter wave radar sensor, and is used to divide the environment into dynamic microenvironment areas according to spatial functions and human activity patterns and define virtual electronic fences; The multimodal sensing module is connected to the cloud management module via the Internet of Things to collect air quality and physiological indicators in each area; The cloud management module is respectively connected with the environment-health coupling assessment module, the dynamic threshold deviation detection module, the spatiotemporal multidimensional trend deduction module, and the regional space purification model module to establish personalized health and wellness air quality dynamic standards; The environment-health coupling assessment module generates an initial regional air-health assessment value through a three-layer neural network structure; the dynamic threshold deviation detection module uses LSTM and attention mechanism to determine whether the assessment value deviates from the standard; the spatiotemporal multidimensional trend deduction module predicts air quality trends and generates target assessment values ​​through the fusion of graph convolutional network and LSTM; the regional space purification model module uses a dual deep Q network to generate purification strategies; The purification equipment linkage execution module receives strategies through the Internet of Things and uses millimeter-wave radar and infrared thermal imaging data to dynamically adjust the direction, speed and range of the purification airflow.

Images