A building door and window indoor monitoring and adjusting intelligent cloud system and adjusting method

Abstract

The application provides a building door and window indoor monitoring and adjusting intelligent cloud system and an adjusting method, comprising an indoor environment precision detection and sensing subsystem, integrated modular detection and sensing indoor environment information and human body health physiological state information, performing indoor environment monitoring and indoor human body health physiological monitoring; a data processing information instruction subsystem, linking a local health decision node, used for processing indoor environment precision detection and sensing subsystem data in real time and issuing control instructions, realizing low-delay privacy calculation; a cloud end processing and scheduling subsystem, used for data storage, big data analysis model training and intelligent control scheduling; an execution control driving subsystem, receiving instructions and controlling door and window execution actions, performing door and window stepless adjustment and emergency unlocking to adjust door and window states; the indoor environment precision detection and sensing subsystem, the data processing information instruction subsystem, the cloud end processing and scheduling subsystem and the execution control driving subsystem are connected through a communication network.

Description

Technical Field

[0001] This invention relates to the field of intelligent cloud-edge precision monitoring and control technology for energy-saving and environmentally friendly buildings, and particularly to an intelligent cloud system and method for monitoring and regulating indoor doors and windows of buildings. Background Technology

[0002] Current building window and door technologies lack environmental data-based adjustment functions or only implement simple opening and closing logic based on rainfall and meteorological data. With the upgrading and intelligent development of building technology, the practicality of building windows and doors is becoming increasingly important for indoor environmental regulation. As the demand for comfortable living standards continues to rise, and with the aging of society, the risk of sudden health emergencies for elderly people living alone and disabled individuals indoors is increasing (such as falls, sleep apnea, and abnormal heart rates). When the life and health of people living alone are at risk, they often find themselves helpless and difficult to be discovered and taken to the hospital in time. The aging population and the trend of not marrying are leading to an increasing number of people living alone, and sometimes even when alone in non-residential areas, there is a risk of accidents, which can easily lead to serious accidents that go unnoticed in time. Buildings require sound insulation and protection against external prying. When doors and windows are closed, it is difficult for outsiders to obtain information about the risk of injury to people and to detect dangerous situations inside the building in a timely manner. Therefore, it is not suitable for modern buildings, age-friendly housing, care facilities for disabled people, and high-end healthy housing. It is necessary to design a building door and window adaptive adjustment system that can monitor vital signs. Most existing smart door and window technologies are based on independent, localized control, relying solely on switches, remote controls, mobile apps, or voice commands. They lack environmental awareness and autonomous decision-making capabilities, and cannot adjust their status based on real-time subtle changes in the environment. A few smart doors and windows possess basic environmental awareness, but their control systems primarily rely on single environmental parameters (indoor temperature, outdoor rainfall) for passive adjustment, or execute simple switching logic based on single outdoor environmental data (rainfall). Furthermore, existing systems suffer from the following technical deficiencies: 1. Disconnect between health monitoring and building control: Existing health monitoring equipment (such as wearable devices and cameras) and door and window control systems operate separately and independently, and cannot automatically adjust indoor ventilation, heat preservation or emergency opening of passages based on real-time human vital signs (such as reduced breathing rate during deep sleep); 2. Limitations of contact-based monitoring: Wearable devices have poor compliance, cannot monitor when not wearing the device, pose a privacy risk to the camera, and cannot accurately monitor sleep stages in dark environments; 3. Lack of health risk-driven emergency safety mechanisms for doors and windows: When people fall or suffer sudden illness, existing doors and windows cannot automatically unlock to form a rescue passage, nor can they adjust the indoor environment according to the patient's vital signs (such as reducing VOC concentration to reduce respiratory burden). 4. The contradiction between privacy protection and real-time response: Health data involves sensitive privacy, and existing technologies lack cloud processing and scheduling systems, posing a risk of data leakage. Furthermore, network latency often leads to delayed responses in safety scenarios (such as calls for help) and emergency scenarios (such as falls). Buildings require sound insulation and protection against external prying, but with doors and windows closed, it is difficult to obtain information on the risk of injury to people and to detect dangerous indoor conditions in a timely manner. They are not suitable for age-friendly housing, care facilities for disabled people, and high-end healthy housing. It is necessary to design a building door and window adaptive adjustment system that can monitor vital signs. Changes in the indoor environment, the occurrence of risks, or the incapacity of individuals living alone cannot be promptly communicated and addressed by others; the existing system also has the following technical problems: Existing technologies are based on independent local control of individual products, relying solely on switches, remote controls, mobile apps, or voice commands for human control. They lack environmental awareness and autonomous decision-making capabilities, and cannot autonomously adjust and control the status of doors and windows based on real-time micro-changes in the environment. Therefore, there is an urgent need to construct a smart cloud system and method for monitoring and regulating indoor doors and windows of buildings, which can partially solve the existing technical problems. Summary of the Invention

[0003] This invention addresses the technical shortcomings of existing intelligent building door and window systems, such as single data collection dimensions, incomplete decision-making logic, weak data storage and processing capabilities, fragmented management and control architecture, and conflicts in multi-objective decision-making. Combining the development trend of "global collaboration, data-driven, intelligent and efficient" in smart cities, smart communities, and smart buildings, this invention provides a smart cloud system for indoor monitoring and adjustment of building doors and windows.

[0004] A smart cloud system for monitoring and regulating indoor spaces of building doors and windows includes: The indoor environment precision detection and sensing subsystem integrates modular detection and sensing of indoor environmental information and human health and physiological status information, and performs indoor environmental monitoring and indoor human health and physiological monitoring. The data processing information command subsystem is linked to the local health decision node. It is used to process data from the indoor environment precision detection and sensing subsystem in real time and issue control commands to achieve low-latency privacy computing. The cloud-based processing and scheduling subsystem is used for data storage, big data analysis model training, and intelligent control and scheduling. The execution control and drive subsystem receives instructions and controls the doors and windows to perform actions, including stepless adjustment and emergency unlocking of the doors and windows. The indoor environment precision detection and sensing subsystem, data processing information command subsystem, cloud processing and scheduling subsystem, and execution control and drive subsystem are connected through a communication network. By combining indoor environmental data collection, an indoor environmental data collection and perception layer is formed, and a multi-layer architecture of edge-cloud is constructed to control and execute the control and drive subsystem. The multimodal millimeter-wave radar group of the human health and physiological monitoring unit and the environmental monitoring sensor group of the indoor environment monitoring unit constitute the indoor environment acquisition and sensing end layer; Data processing information command subsystem: Local health decision node (integrated AI acceleration chip), which performs real-time privacy calculation of vital sign data and issues emergency control commands; Cloud-based: Long-term health record storage, non-sensitive environment optimization model training, and remote operation and maintenance.

[0005] Preferred, The indoor environment precision detection and sensing subsystem includes: an indoor environment monitoring unit and a human health and physiological monitoring unit; The indoor environmental monitoring unit integrates a temperature and humidity sensor group, a VOC sensor, a gas sensor, and a smoke sensor; The human health and physiological monitoring unit integrates a multi-modal millimeter-wave radar array to achieve non-contact monitoring; the multi-modal millimeter-wave radar array includes: presence detection millimeter-wave radar, vital sign monitoring radar, and fall detection radar; The multimodal millimeter-wave radar group achieves data fusion through spatiotemporal synchronization control; Indoor environmental data acquisition and sensing edge layer design: Indoor Environmental Monitoring Unit: This unit integrates multiple high-precision sensors, which are installed inside the indoor control terminal to collect indoor environmental parameters in real time, including but not limited to: temperature sensor (capacitive temperature sensor), humidity sensor (capacitive humidity sensor), VOC sensor (semiconductor air quality detector), gas leak sensor (semiconductor sensor), and smoke sensor (semiconductor smoke detector). Each sensor adopts a modular integrated design, supports hot-swapping and calibration, and the data sampling frequency can be set according to requirements. Indoor Human Health and Physiological Monitoring Unit: Integrates multiple millimeter-wave radar modules and environmental monitoring sensor groups, installed in the indoor control terminal to achieve non-contact real-time monitoring.

[0006] Preferred, The vital signs monitoring radar is used to monitor respiratory rate, heart rate, snoring patterns, and sleep stages. The fall detection radar is used to identify fall events through sudden changes in altitude and attitude angle. The presence detection millimeter-wave radar is used for personnel location, intrusion detection, and micro-motion identification.

[0007] Preferred, The data processing information instruction subsystem is equipped with an intelligent AI model for sleep stage recognition, fall detection, and privacy computing. The privacy-preserving edge computing architecture employs federated learning and cloud encryption mechanisms to address the sensitivity of vital sign data. Data is processed in layers: raw radar echo data is stored at the edge nodes; the cloud receives anonymized control command logs and health risk event markers. Intelligent AI models: Sleep stage recognition model and fall detection model are deployed on edge nodes; An emergency bypass mechanism is adopted: when the network is interrupted, the edge node independently executes emergency control based on the local rule base to maintain offline availability.

[0008] The linkage control logic is based on cloud-based decision-making results, as shown in the following example: Routine adjustment: If indoor VOC concentration is detected to be excessive and human breathing rate is abnormally high, the cloud will instruct the doors and windows to open by 30% and activate the fresh air system (if applicable) until the VOC concentration drops to the safe threshold and the breathing rate returns to normal. Sleep mode: When the millimeter-wave radar for vital signs monitoring detects that the human body has entered a deep sleep state, the cloud command closes the doors and windows to the minimum opening degree (5%) to ensure heat preservation and quietness, and at the same time, the opening degree is finely adjusted according to the indoor temperature to maintain a comfortable environment; Safety alerts: When the human body sensing millimeter-wave radar detects an abnormal intrusion, it commands the doors and windows to be locked and triggers an alarm; when a person falls, the millimeter-wave radar detects a fall and immediately commands the doors and windows to be unlocked, while sending an alert message to the preset emergency contact and triggering the indoor alarm device.

[0009] Preferred, The cloud-based processing and scheduling subsystem includes a multi-level priority decision-making mechanism, including security priority, comfort priority, and energy-saving priority. When the aforementioned security priority is triggered, other priority instructions are blocked, and emergency control is executed. The cloud includes federated learning to aggregate and globally optimize model parameters, while isolating and protecting raw physiological data.

[0010] Preferred, The local health decision-making node integrates an AI chip to enable privacy computing and emergency command issuance; The triggering conditions for the safety priority include at least one of the following: indoor gas leak, excessive smoke, intrusion by personnel, and warning of extreme weather.

[0011] Preferably, the system further includes a health-environment multi-dimensional fusion decision model for dynamically adjusting the weights of health parameters and environmental parameters; When health parameters are abnormal, increase the weight of health parameters and implement respiratory-friendly environmental adjustments.

[0012] Preferably, the building doors and windows receive instructions from a cloud server or indoor control terminal to perform actions, thereby realizing door and window opening adjustment, opening and closing control, and emergency unlocking; supporting stepless opening adjustment from 0-100% and having an emergency unlocking function; In the event of a gas leak or smoke alarm, the system quickly unlocks the nearest doors and windows and activates full ventilation mode; It enables adjustment of door and window opening, control of opening and closing, and emergency unlocking.

[0013] Preferably, the system supports federated learning and cloud encryption mechanisms. The original vital signs data are stored only at the edge nodes, and the cloud only receives desensitized instruction logs. The cloud processing and scheduling subsystem stores long-term health data, non-sensitive model training, remote operation and maintenance, and intelligent decision-making.

[0014] This invention discloses a method for monitoring and regulating the interior of building doors and windows, employing the aforementioned monitoring and regulation method of a smart cloud system for monitoring and regulating the interior of building doors and windows, comprising the following steps: The indoor environment precision detection and sensing subsystem collects indoor environmental data and human vital signs data in real time. Real-time health risk identification and privacy calculations are performed in the data processing information instruction subsystem; The cloud-based scheduling subsystem performs multi-dimensional data fusion and intelligent decision-making to generate control commands. The execution control drive subsystem drives the doors and windows to perform actions such as opening adjustment, opening and closing control, or emergency unlocking according to control commands.

[0015] End-Edge-Cloud Multi-Layer Architecture Design: Cloud servers perform multiple functions, including data storage, big data analysis, and AI-powered intelligent decision-making, as detailed below: Data storage: A distributed database is built to store historical monitoring data (environmental data, human health data) and system execution records (door and window adjustment logs, early warning records). The data retention period can be set, and retrieval by time and parameter type is supported. Basic Rules / Threshold Triggers: Built-in configurable pre-defined rule base, threshold parameter base, and priority rule base form a "priority-rule-threshold" mapping. The rule base contains basic logic for door and window control (e.g., opening ventilation mode when indoor VOC concentration exceeds the standard, immediately opening doors and windows and issuing an early warning for gas leaks, etc.); the threshold parameter base contains preset environmental and safety parameter thresholds based on different building types (residential, commercial, public buildings), building floors, and regional characteristics (e.g., indoor PM2.5 safety threshold for residential buildings is 35μg / m³, gas leak trigger threshold is natural gas concentration ≥0.05%VOL, smoke trigger threshold is smoke concentration ≥0.1mg / m³, for high-rise 34-story residential buildings, windows must be closed when summer wind speed is ≥6 on floors 1-10, closed when summer wind speed is ≥4 on floors 11-20, and closed when summer wind speed is ≥2 on floors 21-34, etc.); the priority rule base clearly defines the triggering conditions and execution logic for three core priority categories: ① Safety Priority (Highest Level): Triggering conditions include external risks (receiving external risk warnings: extreme weather information such as rainstorm warnings and typhoon warnings) and internal risks (indoor gas concentration ≥0.05%VOL, smoke concentration ≥0.1mg / m³, human intrusion detected by human perception sensors, etc.). Once activated, it triggers emergency control commands (such as immediately locking doors and windows and uploading linkage alarm signals) and blocks other priority commands. ② Comfort Priority (Intermediate): Activated when indoor environmental parameters deviate from the comfort threshold (such as temperature <22℃ or >26℃, humidity <40% or >60%), provided there is no safety risk, prioritizing the comfort of indoor living / use. ③ Energy Saving Priority (Regular Level): Activated by default when there are no safety or comfort risks, with the core objective of reducing energy consumption in building ventilation, air conditioning, and other systems. This unit performs priority matching and rule verification on the pre-processed real-time data. If the conditions for triggering a high priority are met, the corresponding instruction is executed directly. If there are priority conflicts or ambiguous parameter ranges (such as a contradiction between comfort requirements and energy saving requirements), the data is pushed to the big data analysis unit.

[0016] Based on historical data, a user health model and an environment adaptation database are constructed to analyze the correlation between vital signs and environmental parameters of different groups (age, health status), such as the optimal temperature and humidity range for elderly people during sleep and the effect of VOC concentration on respiratory rate. The AI ​​model uses massive amounts of historical environmental data, urban instruction data, door and window control records, building feature data, and priority decision labels (including safety priority event records) as training samples. It optimizes model parameters through transfer learning to adapt to building door and window control scenarios. ①Feature encoding: A multi-head attention mechanism is used to extract key features from multi-dimensional data and incorporate them into a priority feature encoding branch, focusing on capturing security priority triggering conditions, multi-priority conflict features, and priority switching timing features; ② Multi-task learning: Simultaneously train four sub-tasks: "risk identification", "optimal control strategy formation", "equipment status adaptation" and "priority dynamic scheduling". Among them, the priority dynamic scheduling sub-task outputs the optimal priority and weight allocation scheme based on real-time data characteristics (especially safety risk characteristics). It adopts an interpretable multi-engine model of "pre-defined rules + big data analysis + AI optimization" to realize multi-level decision-making logic. The pre-defined rules cover national standards (such as indoor formaldehyde concentration ≤0.1mg / m³), health thresholds (such as adult resting heart rate 60-100 beats / minute), and ethical boundary rules (such as not disclosing sensitive health data in non-emergency scenarios); the deep learning model adopts recurrent neural networks (RNN), and the training data comes from compliantly collected environmental and vital sign correlation data of different groups (age, health status).

[0017] The training process includes: data preprocessing (outlier removal and normalization), feature extraction (selecting 12 core features such as temperature, humidity, and respiratory rate), model iteration (≥500 iterations, convergence accuracy ≤0.001), and scenario adaptation optimization. The model can dynamically adjust the decision threshold based on real-time monitoring data, forming a functional synergy between hardware and software features. For example, it can automatically adjust the opening of doors and windows to 30%-40% to improve ventilation for snoring individuals, and immediately trigger door and window unlocking and emergency warnings for fall events. The decision-making process is traceable and explainable, avoiding the "black box" problem.

[0018] Multimodal millimeter-wave radar integrated design This invention integrates multiple radar modes within a single indoor control terminal and achieves data fusion through spatiotemporal synchronization control: 1. Existing millimeter-wave radar (60GHz, detection range 0.3-10m): Functions: Personnel location, intrusion detection, and micro-motion recognition; Linked with doors and windows: Triggers door and window locking and local alarm when unauthorized intrusion is detected; 2. Vital signs monitoring radar (60GHz, chest cavity detection range 0.4-1.5m): Functions: respiratory rate (accuracy ±1 breaths / minute), heart rate (accuracy ±2 breaths / minute), snoring pattern recognition, sleep stage classification (based on a threshold determination model of body movement and respiratory rhythm); Linked with doors and windows: Detects deep sleep (body movement <3 times / hour + respiratory rate 12-16 breaths / minute) → instructs doors and windows to close to 5% and activates silent mode; Detects snoring (frequency >40 times / hour) → opens doors and windows by 30% to improve ventilation; 3. Fall detection radar (60GHz, vertical accuracy ±5cm): Function: Fall detection is achieved by detecting sudden changes in height (>50cm / second) and changes in posture angle (horizontal angle <30° for >2 seconds), with a false alarm rate of <0.5%. Linked with doors and windows: Immediately unlock the nearest doors and windows (to create a rescue passage) + open all ventilation (reduce VOC to <0.05mg / m³ to reduce the patient's respiratory burden) + send an encrypted alert with location coordinates to the cloud.

[0019] Health-Environment Multidimensional Integrated Decision-Making Model Construct a fusion decision model of health risk coefficient (H) and environmental parameters (E): 1. Dynamic adjustment of health risk weights: Normal state: Health parameters account for 30% of the weight, and environmental parameters account for 70% of the weight. Abnormal conditions (such as respiratory rate >25 breaths / minute or <8 breaths / minute): The weight of health parameters is automatically increased to 80%, and respiratory-friendly environmental adjustments are prioritized (such as reducing the VOC threshold from 0.1 mg / m³ to 0.05 mg / m³, and increasing ventilation). 2. Sleep Stage - Door and Window Opening Mapping Table (Non-linear Mapping): Awake period: Door and window opening is set manually by the user or determined by environmental parameters; Light sleep period (N1 / N2): Open doors and windows to 20% to maintain background ventilation; Deep sleep period (N3): Door and window opening degree 5%, prioritize heat preservation and noise reduction; REM period: Doors and windows are opened to 40% to compensate for the increased oxygen demand caused by the increased metabolic rate; 3. Fall Emergency Response Protocol: T0 (Fall Detection, <1s): Edge nodes immediately unlock doors and windows; T0+5s: If the vital signs radar detects a sudden drop in respiratory rate (<6 breaths / minute), it will automatically turn on maximum ventilation (100% opening) and reduce indoor VOCs; T0+30s: The emergency contact is notified via the cloud to upload vital signs data from the 30 seconds prior to the fall. Privacy-preserving edge computing architecture To address the sensitivity of vital sign data, a federated learning + cloud encryption mechanism is employed. 1. Data layered processing: Raw radar echo data (including chest micro-motion waveforms) is stored only at the edge nodes and is not uploaded to the cloud; The cloud only receives anonymized control command logs (such as "deep sleep - close doors and windows") and health risk event markers (such as "fall - handled"). 2. Intelligent AI Model: The sleep stage recognition model and fall detection model are deployed on edge nodes (TensorRT optimized), with inference latency <50ms; The cloud only performs global optimization of model parameters (federated learning aggregation), isolating and protecting raw physiological data; 3. Emergency bypass mechanism: When the network is interrupted, edge nodes independently execute emergency controls (such as fall unlock) based on local rule bases to maintain offline availability.

[0020] Specialized fit for disabled individuals 1. Long-term bed rest mode: Vital signs radar continuously monitors the risk of pressure sores (body position remains unchanged for a long time), and adjusts indoor humidity (maintaining 50%-60%) in combination with the opening of doors and windows. 2. Medication Reminder Linkage: Upon detecting a patient who is awake and still (in a pre-medication posture) → automatically open doors and windows for 10 minutes to reduce indoor pollutant concentration and decrease the metabolic burden on the medication. Compared to existing technologies, this invention offers the following advantages: Comprehensive and highly integrated monitoring dimensions: For the first time, multi-sensor environmental monitoring and various millimeter-wave radar health monitoring are integrated into the same terminal, achieving full-scenario coverage of environment, safety and health, and solving the problem of disconnection in existing systems; Breaking through the limitations of traditional single-product control, achieving end-to-end intelligent collaboration: Replacing traditional remote controls, control panels, and voice with a dedicated integrated control system, enabling real-time linkage and autonomous decision-making between indoor and outdoor environments and human health parameters. Driven by a dual engine of AI and pre-defined rules, it promotes adaptive adjustment of doors and windows, completely solving the industry pain point of independent operation and lack of collaboration of single products; At the same time, it integrates AI ethical control and open model design to solve the black box problem, and deeply collaborates with the technical features of models, sensors, door and window actions, communication modules, etc. Enhancing the accuracy of intelligent decision-making and risk prevention capabilities: Through the synergy of multi-level computing logic, dynamic priority decision-making mechanism, and public meteorological data fusion, it accurately identifies and predicts safety priority risk scenarios in advance (such as activating safety priority 30 minutes in advance by combining public meteorological precipitation forecasts), shortens the delay in handling safety incidents, and improves the accuracy of risk prediction; at the same time, it solves the problem of multi-objective decision-making conflicts, improves the decision-making accuracy in complex scenarios, and is significantly better than traditional fixed logic control.

[0021] Achieve global collaborative management and control: With a cloud platform as the core and a priority decision-making mechanism, it breaks the traditional independent operation mode of door and window equipment and realizes community-level and city-level collaborative scheduling of doors and windows (such as activating the safety priority across the entire area and simultaneously closing and locking doors and windows when a red rainstorm warning is issued), supporting the construction goal of "global perception and collaborative control" of smart cities.

[0022] Precise and efficient linkage control: Directly links with core building components such as doors and windows, dynamically adjusts the environment based on human health status, and improves living comfort and safety; Enhance system stability and scalability: The cloud-based centralized architecture enables real-time data backup, remote operation and maintenance, and module upgrades, avoiding insufficient computing power and failure risks of local devices; the hybrid storage architecture can support massive data storage, and the standardized interface design facilitates the integration of new sensor types and management systems, with strong scalability.

[0023] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in this patent application.

[0024] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0025] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the architecture of a smart cloud system for monitoring and regulating indoor doors and windows in a building, as described in an embodiment of the present invention. Figure 2 This is a design drawing of a smart cloud system for monitoring and regulating indoor spaces of building doors and windows according to the present invention; Figure 3 This is a diagram illustrating a smart cloud system terminal and APP for indoor monitoring and adjustment of building doors and windows according to the present invention. Detailed Implementation

[0026] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0027] As shown in the figure, this embodiment of the invention provides a smart cloud system for monitoring and regulating indoor spaces of building doors and windows, specifically including: Taking the bedroom of an elderly person living alone as an example: The indoor environment precision detection and sensing subsystem integrates modular detection and sensing of indoor environmental information and human health and physiological status information, and performs indoor environmental monitoring and indoor human health and physiological monitoring. Vital signs monitoring radar monitors rest areas; human presence radar monitors entry and exit areas; fall detection radar monitors fall risk areas; rest areas include sleeping areas or sitting / lying rest areas; entry and exit areas include indoor door entry and exit areas or indoor passageway entry and exit areas; fall risk areas include indoor slippery areas or areas with obstacles that are prone to falling. Indoor control terminals are deployed and installed in indoor scenarios such as bedrooms, living rooms, bathrooms, and kitchens. The terminals integrate indoor environmental monitoring units, human health and physiological monitoring units, and control terminals. Each terminal covers the room space corresponding to the doors and windows, and monitors various environmental and personnel status parameters such as indoor environmental conditions, indoor human health status, and indoor personnel posture in real time. All multimodal millimeter-wave radar groups and environmental monitoring sensor groups are powered by a control circuit board power supply group; the control circuit board integrates a communication module to transmit signals to the edge computing gateway (with a built-in NPU chip). The data processing information command subsystem is linked to the local health decision node. It is used to process data from the indoor environment precision detection and sensing subsystem in real time and issue control commands to achieve low-latency privacy computing. The cloud-based processing and scheduling subsystem is used for data storage, big data analysis model training, and intelligent control and scheduling. The execution control and drive subsystem receives instructions and controls the doors and windows to perform actions, including stepless adjustment and emergency unlocking of the doors and windows. The indoor environment precision detection and sensing subsystem, data processing information command subsystem, cloud processing and scheduling subsystem, and execution control and drive subsystem are connected through a communication network. Scenario 1: Sleep Breathing Optimization At 22:00, the millimeter-wave radar module detected that the elderly person had gone to bed and had a breathing rate of 16 breaths per minute (awake), and automatically entered sleep mode; At 22:30, the respiratory rate drops to 12 breaths / minute and body movement decreases to below the body movement threshold. The threshold determination model determines that the person has entered deep sleep. The body movement threshold includes 1 breath / hour or 2 breaths / hour. Edge node instructions: Close bedroom doors and windows to 5% opening (leaving only a small gap), and keep living room doors and windows at 20% opening to form an air buffer zone; at 03:00, if an increase in snoring frequency (>50 times / hour) is detected, automatically open bedroom doors and windows to 35% to introduce clean air from the living room buffer zone. After 10 minutes, the snoring frequency decreases to 20 times / hour, and the opening is restored to 5%.

[0028] One embodiment, The indoor human health and physiological monitoring unit specifically includes: (1) Human body sensing millimeter-wave radar: adopts a 60GHz millimeter-wave indoor human presence sensing and tracking radar subsystem, with a detection range of 0.3-10 meters, a detection angle of ±60° azimuth and ±40° (-3dB) degrees, a distance measurement accuracy of ≤0.1m, a presence detection accuracy of 98% (Typ.), a response time of ≤100ms, and identifies human body contours and movement trajectories through the micro-Doppler effect to distinguish between humans and pets. When unauthorized personnel intrusion is detected, an early warning is triggered, with a false alarm rate of ≤0.5%. (2) Millimeter-wave radar for vital signs monitoring: A 60GHz millimeter-wave vital signs (respiration and heart rate) monitoring radar subsystem is used to capture sub-millimeter-level movements of the human chest and extract respiratory frequency (respiratory and heart rate detection distance (chest cavity) 0.4~1.5m, respiratory measurement accuracy 90% (Typ.), heart rate measurement accuracy 90% (Typ.), recognition accuracy 90% (Typ.)), heart rate (monitoring range 40-180 beats / minute, accuracy ±2 beats / minute), snoring pattern (identifying snoring duration and frequency), and combine with movement status to determine sleep stage (light sleep, deep sleep, REM sleep); (3) Millimeter-wave radar for falling people: It adopts a 60GHz millimeter-wave fall detection and vital sign monitoring radar subsystem with a detection range of 0.3~10m, a detection angle of ±60° azimuth and ±40° (-3dB) degrees of pitch, and a fall recognition accuracy of 95% (Typ.). It supports all-weather operation and identifies fall actions by monitoring changes in human height, motion acceleration and posture characteristics (response time ≤1s). It is especially suitable for groups such as the elderly with mobility difficulties and disabled people, and can distinguish falls from normal squatting, sitting and lying down actions.

[0029] The indoor environment precision detection and sensing subsystem includes: an indoor environment monitoring unit and a human health and physiological monitoring unit; The indoor environmental monitoring unit integrates a temperature and humidity sensor group, a VOC sensor, a gas sensor, and a smoke sensor; The human health and physiological monitoring unit integrates a multi-modal millimeter-wave radar array to achieve non-contact monitoring; the multi-modal millimeter-wave radar array includes: presence detection millimeter-wave radar, vital sign monitoring radar, and fall detection radar; The multimodal millimeter-wave radar group achieves data fusion through spatiotemporal synchronization control; Examples of nighttime fall rescue: At 02:00, the elderly person got up at night, and the fall detection radar detected a sudden change in height (from 1.6m to 0.4m) and horizontal posture; Within 1 second, the edge node triggers the following: unlocks the bedroom door (creating a rescue passage), opens all doors and windows to 100% (for ventilation and VOC reduction), and turns on soft lighting; if the vital signs radar detects a normal respiratory rate (14 breaths / minute), it is determined to be a minor fall, and the edge node asks "Do you need help?" via local voice; if there is no response within 30 seconds or the respiratory rate is <8 breaths / minute, it automatically activates a local alarm and sends a remote alarm message containing accurate time and precise location coordinates (bedroom / living room / bathroom) to the user's mobile terminal (mobile APP).

[0030] In one example, respiratory load is reduced: The vital signs monitoring radar is used to monitor respiratory rate, heart rate, snoring patterns, and sleep stages. The fall detection radar is used to identify fall events through sudden changes in altitude and attitude angle. The presence detection millimeter-wave radar is used for personnel positioning, intrusion detection, and micro-motion identification; When the VOC concentration is 0.08 mg / m³ (lower than the national standard of 0.1 mg / m³), doors and windows should be kept closed under normal circumstances. However, the vital signs radar detected that the elderly person's respiratory rate had increased to 22 breaths per minute (anxiety or mild hypoxia). The health-environment multi-dimensional fusion decision-making model determined that although VOC levels were not exceeded, based on the increased health weight, the door and window opening was instructed to be increased by 25% to reduce VOC to 0.04 mg / m³, and the respiratory rate recovered to 16 breaths / minute after 15 minutes; Model implementation details: Sleep stage classification model: Based on a one-dimensional CNN network, the input is a 30-second sliding window of respiratory waveforms and body movement counts, and the output is a four-class classification of wakefulness / light sleep / deep sleep / REM sleep, with an accuracy of >92%; Fall detection model: A point cloud-based height clustering model is used, combined with Kalman filtering to track the vertical velocity of the human centroid. An early warning is triggered when the velocity is >1.5m / s and the final height is <0.8m. Privacy encryption: Edge nodes use TEE (Trusted Execution Environment) to store raw radar data, so that physiological characteristics cannot be extracted even if the device is physically disassembled; The data processing information instruction subsystem is equipped with an intelligent AI model for sleep stage recognition, fall detection, and privacy computing. The privacy-preserving edge computing architecture employs federated learning and cloud encryption mechanisms to address the sensitivity of vital sign data. Data is processed in layers: raw radar echo data is stored at the edge nodes; the cloud receives anonymized control command logs and health risk event markers. Intelligent AI models: Sleep stage recognition model and fall detection model are deployed on edge nodes; An emergency bypass mechanism is adopted: when the network is interrupted, the edge node independently executes emergency control based on the local rule base to maintain offline availability; One example, a security priority scenario: Indoor gas concentration (Data A), smoke concentration (Data B), occupant intrusion (Data C), and connected public meteorological data (air pressure 1005 hPa and forecast of short-term heavy precipitation within 1 hour, Data D) are treated as four independent data sources. The weight of safety priority data is set at 0.6 (A, B, and C combined), the weight of public meteorological data is 0.3, and the weight of other data is 0.1. Fusion is achieved through the following steps: ① Determine the identification framework θ = {safe, low risk, high risk}; ② Calculate the basic probability allocation (BP) of each data source. A), such as the BPA of data A (indoor gas concentration = 0.06% VOL) ​​being {Safe: 0.1, Low risk: 0.2, High risk: 0.7}, and the BPA of data D (public weather forecast of short-term heavy precipitation) being {Safe: 0.05, Low risk: 0.25, High risk: 0.7}; ③ Introducing a conflict coefficient correction formula to solve the data conflict problem of traditional DS theory, after correction, the BPA of the four data sources are merged, and the fusion result is output (such as the high-risk BPA after fusion = 0.92), triggering safety priority decision; The cloud-based processing and scheduling subsystem includes a multi-level priority decision-making mechanism, including security priority, comfort priority, and energy-saving priority. When the aforementioned security priority is triggered, other priority instructions are blocked, and emergency control is executed. Cloud-based systems, including federated learning aggregation, only perform global optimization of model parameters, isolating and protecting raw physiological data.

[0031] In one embodiment, a local health decision-making node integrates an AI chip to enable privacy-preserving computation and emergency command issuance; The triggering conditions for the safety priority include at least one of the following: indoor gas leak, excessive smoke, intrusion by personnel, and warning of extreme weather.

[0032] In one embodiment, the system further includes a health-environment multi-dimensional fusion decision model for dynamically adjusting the weights of health parameters and environmental parameters; When health parameters are abnormal, increase the weight of health parameters and implement respiratory-friendly environmental regulation; Model Details: The linkage analysis and decision-making model adopts a multi-engine mode of "pre-defined rule filtering + big data analysis + AI multi-dimensional real-time fusion". The specific decision-making method is as follows: First, real-time data preprocessing: indoor and outdoor environmental parameters and human health parameters are encrypted and transmitted to the cloud. Outlier removal and normalization are performed simultaneously to maintain data validity, with a processing latency of ≤100ms. Second, pre-defined rule real-time filtering: based on national standards, health thresholds, and ethical rules, it quickly responds to emergency scenarios (such as immediately triggering door and window opening + local alarm + remote alarm when gas concentration is ≥0.1%, and triggering door and window half-opening + audible and visual alarm when smoke concentration exceeds the standard), with a response time of ≤200ms, avoiding reliance on manual intervention. The third step involves big data analysis and AI-driven real-time decision-making. It inputs all real-time indoor and outdoor parameters (indoor temperature and humidity, VOC concentration, etc.; outdoor PM2.5, wind speed, etc.; human respiratory rate, heart rate, posture, etc.), and calculates dynamic weights (60% for health parameters, 30% for indoor environmental parameters, and 10% for outdoor environmental parameters) using a trained RNN model. It outputs real-time door and window opening degree (accuracy ±5%), opening / closing status, and warning level, with a decision cycle ≤500ms. The fourth step involves instruction feedback and optimization. Door and window actions are fed back in real-time, execution logs are recorded in the cloud and incorporated into big data analysis, continuously iterating AI model parameters and optimizing decision accuracy. The entire decision-making process is traceable and interpretable, replacing traditional passive control while achieving proactive adaptive adjustment.

[0033] Example of using a smart cloud platform as the core software control center for the entire smart door and window system: It integrates a multi-level dynamic priority decision-making mechanism for safety, comfort, and energy saving (clearly defining safety priorities to cover multiple indoor risk scenarios), incorporates a standardized public meteorological data connection module, and realizes real-time fusion analysis of public meteorological data and multi-dimensional data, replacing the traditional decentralized local control architecture and improving the system's coordination, predictive ability, and intelligence level; Multi-level computational logic: Create a computational mechanism that deeply integrates "basic rules / threshold triggering - big data analysis - AI model operation" with priority dynamic scheduling. This ensures immediate response in safety priority scenarios and achieves optimized balance of multiple objectives (comfort, energy saving) in normal scenarios, thus overcoming the limitations of a single computational logic. Specialized Big Data and AI Models: For building door and window control scenarios, we optimize the multi-source data fusion model (introducing priority weights) and time series prediction model, and build an AI model with priority coding branches to achieve deep adaptation of environmental data, urban instructions, safety monitoring data and priority decision-making.

[0034] Model Intelligent Training: Structured health-environment monitoring data (such as respiratory rate, temperature, humidity, and regular serialized data) is input into branch N1 of the intelligent AI model's recurrent neural network (RNN) to obtain the first feature output by the intelligent AI model; unstructured health-environment monitoring data (such as body movement, posture, falls, and data with irregular structure or low integrity) is input into branch N2 of the intelligent AI model's recurrent neural network (RNN) to obtain the second feature output by the intelligent AI model; the first and second features output by the intelligent AI model are fused to obtain a deep fusion feature; this deep fusion feature is then used as input to the decision-making model, based on existing decision execution data. The training process involves: using the preset accuracy of the decision as the training label; performing a softmax function operation on the output layer of the recurrent neural network (RNN) at each time step; using cross-entropy to lose the error between the model output and the training label; iteratively calculating the hidden states in the hidden layers, with the output at the Nth time step determined by the feature sequence number; training the feature sequence of the decision execution data; the loss at the Nth time step is calculated by the probability distribution of the next decision execution data feature, such as a binomial distribution function; determining the next decision execution data feature based on the feature sequence and the label of the time step; and continuing until the preset accuracy of the decision is reached.

[0035] In one embodiment, the building doors and windows perform actions including receiving instructions from a cloud server or an indoor control terminal to adjust the opening degree of the doors and windows, control their opening and closing, and enable emergency unlocking; they support stepless adjustment of the opening degree from 0-100% and have an emergency unlocking function; In the event of a gas leak or smoke alarm, the system quickly unlocks the nearest doors and windows and activates full ventilation mode.

[0036] In one embodiment, the system supports federated learning and cloud encryption mechanisms. Raw vital sign data is stored only at edge nodes, and the cloud only receives desensitized instruction logs. The cloud processing and scheduling subsystem stores long-term health data, non-sensitive model training, remote operation and maintenance, and intelligent decision-making. Cross-platform collaboration capability: Through standardized interface design, it achieves seamless integration with smart city, smart community, smart building related systems and other smart home systems. It can receive priority scheduling instructions from higher-level systems and support the collaborative operation of building terminals and city-level management and control systems.

[0037] This invention discloses a method for monitoring and regulating the interior of building doors and windows, employing the aforementioned monitoring and regulation method of a smart cloud system for monitoring and regulating the interior of building doors and windows, comprising the following steps: The indoor environment precision detection and sensing subsystem collects indoor environmental data and human vital signs data in real time. Real-time health risk identification and privacy calculations are performed in the data processing information instruction subsystem; The cloud-based scheduling subsystem performs multi-dimensional data fusion and intelligent decision-making to generate control commands. The execution control drive subsystem drives the doors and windows to perform actions such as opening adjustment, opening and closing control, or emergency unlocking according to control commands; Core hardware integration method: The indoor control terminal integrates multi-dimensional indoor environmental monitoring units with various millimeter-wave radars to achieve the integration of non-contact health monitoring and environmental monitoring. Real-time linkage decision-making model and method: Integrating pre-made rules, threshold models and interpretable RNN deep learning models, it clarifies the real-time decision-making method of the whole process of "data preprocessing-rule filtering-AI decision-instruction feedback", realizes dynamic weighted fusion analysis of indoor and outdoor environmental parameters and human vital signs parameters, outputs door and window adjustment instructions, replaces traditional passive operation, and forms functional synergy between model features and technical features. Data security mechanism: A monitoring data transmission and storage scheme based on AES-256 encryption and TLS 1.3 protocol to ensure privacy and security; Scenario-based control logic: Adaptive adjustment strategies for doors and windows in scenarios such as sleep, falls, and abnormal intrusions, especially suitable for the elderly and disabled people; Cloud-edge collaborative architecture: A collaborative design of cloud-based big data analytics and terminal local caching decisions to improve system stability and response speed.

[0038] This invention constructs a building door and window management system with a smart cloud platform at its core, integrating a dynamic priority intelligent autonomous decision-making mechanism for safety, comfort, and energy conservation. It clearly defines safety priorities, encompassing external risks such as city-level extreme weather warnings and internal risks such as indoor gas leaks, smoke monitoring, and intrusion. Through deep integration of "basic rules / thresholds, big data analysis, and AI model operation" with dynamic priority scheduling, it achieves progressive analysis of multi-dimensional data, precise resolution of multi-objective conflicts, and intelligent decision-making prioritizing safety. This addresses the technical pain points of traditional systems, such as weak decision-making capabilities, poor coordination, and insufficient safety control. Furthermore, through standardized interface design and cross-platform collaboration capabilities, it adapts to the development needs of smart cities, smart communities, and smart buildings.

[0039] To address the technical shortcomings of existing technologies that separate vital sign monitoring from building door and window control, this invention achieves the following: An integrated multimodal millimeter-wave radar and real-time environmental sensing adaptive adjustment control system for building doors and windows has been developed. 1. Closed-loop integration of non-contact vital signs, real-time indoor environment perception and door and window control: Through the real-time linkage of multiple millimeter-wave radar modes (human presence perception, vital signs monitoring, fall detection) and multiple environmental monitoring sensor modules (temperature and humidity, VOC concentration, gas, smoke) with door and window actions, an active health protection regulation and control system of "health monitoring - risk identification - environmental regulation - emergency access" is constructed. 2. Sleep Stage Adaptive Environmental Adjustment: Based on the sleep stage (light sleep, deep sleep, REM sleep) identified by millimeter-wave radar and combined with the real-time monitoring of the indoor environment by the indoor environment sensing module, the opening degree of doors and windows is dynamically adjusted (e.g., maintaining a 5% opening to ensure quiet during deep sleep and increasing ventilation during REM sleep), replacing traditional timed control; 3. Health risk-driven emergency door and window control: When abnormalities such as falls or breathing apnea are detected, doors and windows are automatically unlocked to form a rescue passage, and indoor environmental parameters are adjusted accordingly (such as opening ventilation to reduce VOCs to alleviate respiratory burden). 4. Privacy-secure edge-cloud processing and scheduling system collaborative architecture: Sensitive vital sign data is processed in real time at local edge nodes (latency <100ms), and only de-identified control commands are uploaded to the cloud processing and scheduling system, balancing privacy protection and intelligent decision-making; 5. Deep integration of health monitoring and building control: For the first time, adaptive adjustment of doors and windows based on vital signs parameters such as indoor environmental conditions, sleep stages, and fall risk has been achieved, filling the technological gap in the field of "healthy buildings"; 6. Non-contact privacy protection: Millimeter-wave radar can penetrate bedding to monitor sleep without a camera, avoiding privacy leaks and making it suitable for private spaces such as bedrooms; 7. Improved emergency response speed: The edge computing architecture reduces the delay from fall detection to door / window unlocking to within 1 second, which is much faster than the cloud processing mode (usually 3-5 seconds), buying more time for rescue. 8. Respiratory-friendly environmental control: For patients with respiratory diseases (such as COPD and asthma), when abnormal respiratory rate is detected, the VOC threshold is automatically tightened and ventilation is enhanced or the fresh air system and air purification system are activated to purify indoor air, thereby achieving therapeutic environmental regulation; 9. Advantages of age-friendly modifications: No need for the elderly to wear devices, no change in living habits, and the risk of the elderly getting cold at night and not being able to be rescued after a fall is reduced by actively adjusting doors and windows; In addition, the present invention also has the following beneficial effects, which are in line with the development trend of deep integration of smart cities, smart communities, smart buildings, and healthy living with smart homes: Comprehensive and highly integrated monitoring dimensions: For the first time, multi-sensor environmental monitoring and various millimeter-wave radar health monitoring are integrated into the same terminal, achieving full-scenario coverage of environment, safety and health, and solving the problem of disconnection in existing systems; Breaking through the limitations of traditional single-product control, achieving end-to-end intelligent collaboration: Replacing traditional remote controls, control panels, and voice with a dedicated integrated control system, enabling real-time linkage and autonomous decision-making between indoor and outdoor environments and human health parameters. Driven by a dual engine of AI and pre-defined rules, it promotes adaptive adjustment of doors and windows, completely solving the industry pain point of independent operation and lack of collaboration of single products; At the same time, it integrates AI ethical control and open model design to solve the black box problem, and deeply collaborates with the technical features of models, sensors, door and window actions, communication modules, etc. Enhancing the accuracy of intelligent decision-making and risk prevention capabilities: Through the synergy of multi-level computing logic, dynamic priority decision-making mechanism, and public meteorological data fusion, it accurately identifies and predicts safety priority risk scenarios in advance (such as activating safety priority 30 minutes in advance by combining public meteorological precipitation forecasts), shortens the delay in handling safety incidents, and improves the accuracy of risk prediction; at the same time, it solves the problem of multi-objective decision-making conflicts, improves the decision-making accuracy in complex scenarios, and is significantly better than traditional fixed logic control.

[0040] Achieve global collaborative management and control: With a cloud platform as the core and a priority decision-making mechanism, it breaks the traditional independent operation mode of door and window equipment and realizes community-level and city-level collaborative scheduling of doors and windows (such as activating the safety priority across the entire area and simultaneously closing and locking doors and windows when a red rainstorm warning is issued), supporting the construction goal of "global perception and collaborative control" of smart cities.

[0041] Precise and efficient linkage control: Directly links with core building components such as doors and windows, dynamically adjusts the environment based on human health status, and improves living comfort and safety; Enhance system stability and scalability: The cloud-based centralized architecture enables real-time data backup, remote operation and maintenance, and module upgrades, avoiding insufficient computing power and failure risks of local devices; the hybrid storage architecture can support massive data storage, and the standardized interface design facilitates the integration of new sensor types and management systems, with strong scalability.

[0042] Data security is ensured by end-to-end encryption to meet privacy protection requirements, while also supporting edge computing to improve system stability. High feasibility: It adopts mature millimeter-wave radar modules and wireless communication protocols, and its modular design facilitates installation and expansion. The cost is controllable and it is easy to scale up and promote. Achieving multi-objective optimization balance: Through priority dynamic scheduling, while ensuring safety, the system maximizes the balance between comfort and energy conservation needs, reduces the energy consumption of building equipment systems, and improves the indoor comfort compliance rate.

[0043] The weighting coefficients and thresholds in the above model are only values ​​of a preferred embodiment of the present invention. In practical applications, those skilled in the art can make linear or nonlinear fine adjustments based on indoor space volume, user physical differences, and sensor accuracy, all of which fall within the protection scope of the present invention.

[0044] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A smart cloud system for monitoring and regulating indoor spaces of building doors and windows, characterized in that, include: The indoor environment precision detection and sensing subsystem integrates modular detection and sensing of indoor environmental information and human health and physiological status information, and performs indoor environmental monitoring and indoor human health and physiological monitoring. The data processing information command subsystem is linked to the local health decision node. It is used to process data from the indoor environment precision detection and sensing subsystem in real time and issue control commands to achieve low-latency privacy computing. The cloud-based processing and scheduling subsystem is used for data storage, big data analysis model training, and intelligent control and scheduling. The execution control and drive subsystem receives instructions and controls the doors and windows to perform actions, including stepless adjustment and emergency unlocking of the doors and windows. The indoor environment precision detection and sensing subsystem, data processing information command subsystem, cloud processing scheduling subsystem, and execution control drive subsystem are connected through a communication network.

2. The intelligent cloud system for monitoring and regulating indoor spaces of building doors and windows according to claim 1, characterized in that, The indoor environment precision detection and sensing subsystem includes: an indoor environment monitoring unit and a human health and physiological monitoring unit; The indoor environmental monitoring unit integrates a temperature and humidity sensor group, a VOC sensor, a gas sensor, and a smoke sensor; The human health and physiological monitoring unit integrates a multi-modal millimeter-wave radar array to achieve non-contact monitoring; the multi-modal millimeter-wave radar array includes: presence detection millimeter-wave radar, vital sign monitoring radar, and fall detection radar; The multimodal millimeter-wave radar group achieves data fusion through spatiotemporal synchronization control.

3. The intelligent cloud system for monitoring and regulating indoor spaces of building doors and windows according to claim 2, characterized in that, The vital signs monitoring radar is used to monitor respiratory rate, heart rate, snoring patterns, and sleep stages. The fall detection radar is used to identify fall events through sudden changes in altitude and attitude angle. The presence detection millimeter-wave radar is used for personnel location, intrusion detection, and micro-motion identification.

4. The intelligent cloud system for monitoring and regulating indoor spaces of building doors and windows according to claim 1, characterized in that, The data processing information instruction subsystem is equipped with an intelligent AI model for sleep stage recognition, fall detection, and privacy computing. The privacy-preserving edge computing architecture employs federated learning and cloud encryption mechanisms to address the sensitivity of vital sign data. Data is processed in layers: raw radar echo data is stored at the edge nodes; the cloud receives anonymized control command logs and health risk event markers. Intelligent AI models: Sleep stage recognition model and fall detection model are deployed on edge nodes; An emergency bypass mechanism is adopted: when the network is interrupted, the edge node independently executes emergency control based on the local rule base to maintain offline availability.

5. The intelligent cloud system for monitoring and regulating indoor spaces of building doors and windows according to claim 1, characterized in that, The cloud-based processing and scheduling subsystem includes a multi-level priority decision-making mechanism, including security priority, comfort priority, and energy-saving priority. When the aforementioned security priority is triggered, other priority instructions are blocked, and emergency control is executed. The cloud includes federated learning to aggregate and globally optimize model parameters, while isolating and protecting raw physiological data.

6. The intelligent cloud system for monitoring and regulating indoor spaces of building doors and windows according to claim 5, characterized in that, The local health decision-making node integrates an AI chip to enable privacy computing and emergency command issuance; The triggering conditions for the safety priority include at least one of the following: indoor gas leak, excessive smoke, intrusion by personnel, and warning of extreme weather.

7. The intelligent cloud system for monitoring and regulating indoor spaces of building doors and windows according to claim 1, characterized in that, The system also includes a health-environment multi-dimensional fusion decision model, used to dynamically adjust the weights of health parameters and environmental parameters; When health parameters are abnormal, increase the weight of health parameters and implement respiratory-friendly environmental adjustments.

8. The intelligent cloud system for monitoring and regulating indoor spaces of building doors and windows according to claim 1, characterized in that, The building doors and windows receive instructions from the cloud server or indoor control terminal to perform actions, realize door and window opening adjustment, opening and closing control, and emergency unlocking; support 0-100% stepless opening adjustment, and have an emergency unlocking function; In the event of a gas leak or smoke alarm, the system quickly unlocks the nearest doors and windows and activates full ventilation mode.

9. A smart cloud system for monitoring and regulating indoor spaces of building doors and windows according to claim 1, characterized in that, The system supports federated learning and cloud encryption mechanisms. Raw vital sign data is stored only at edge nodes, and the cloud only receives desensitized instruction logs. The cloud processing and scheduling subsystem stores long-term health data, non-sensitive model training, remote operation and maintenance, and intelligent decision-making.

10. A method for monitoring and adjusting the interior of building doors and windows, employing the monitoring and adjustment method of any one of the intelligent cloud systems for monitoring and adjusting the interior of building doors and windows according to claims 1-9, characterized in that, Includes the following steps: The indoor environment precision detection and sensing subsystem collects indoor environmental data and human vital signs data in real time. Real-time health risk identification and privacy calculations are performed in the data processing information instruction subsystem; The cloud-based scheduling subsystem performs multi-dimensional data fusion and intelligent decision-making to generate control commands. The execution control drive subsystem drives the doors and windows to perform actions such as opening adjustment, opening and closing control, or emergency unlocking according to control commands.

Images