A multi-sensor fusion environment perception control method and system

Abstract

The present application belongs to the technical field of internet of things intelligent sensing and building energy saving control, and specifically relates to a control method and system for multi-sensor fusion environment sensing, wherein the method comprises the following steps: flexibly configuring millimeter wave radar, carbon dioxide, temperature and humidity and other kinds of sensor modules according to scene requirements; calculating the comprehensive probability of human existence based on multi-source data such as radar micro-motion characteristics and carbon dioxide concentration change rate, and realizing human spatial area positioning and vital sign classification in combination with multi-panel triangular positioning. The present application integrates modular multi-sensor and switch panel, utilizes existing 86 bottom box power supply and installation position, eliminates ceiling hole and independent wiring, and reduces the cost of single room transformation by more than 60%; through multi-sensor fusion, the accuracy of stationary human body detection is improved to more than 98%, and partition fine control is realized, with energy saving of 15%-30%, which can be widely applied to intelligent hotel, smart home, intelligent office and intelligent elderly care scenes.

Description

Technical Field

[0001] This invention relates to the field of IoT intelligent sensing and building energy-saving control technology, and in particular to a control method and system for multi-sensor fusion environmental sensing. Background Technology

[0002] In scenarios such as hotels, homes, and offices, accurately determining whether someone is in a space and in which area they are is fundamental to achieving cardless power access and zoned energy conservation. Current technologies primarily employ the method of independently installing infrared, microwave, or millimeter-wave radar sensors on the ceiling, but this approach has the following drawbacks: The physical separation of sensors from switch panels necessitates independent wiring, drilling, and debugging at each location; data from multiple sensors cannot be fused; power and signal lines need to be laid separately in the ceiling, resulting in high renovation costs for old buildings; single sensors have a high failure rate in detecting stationary human bodies; data from each sensor is fragmented and lacks the ability to fuse and make judgments; sensor configurations are fixed and cannot flexibly adapt to different scenario requirements, from economical to high-end. Summary of the Invention

[0003] This invention proposes a control method and system for multi-sensor fusion environmental perception, aiming to solve the comprehensive technical problems existing in the current intelligent building human body perception technology, such as the need for independent wiring of sensors leading to high transformation costs, the existence of detection blind spots and misjudgments by single sensors, and the inability to adapt to the diverse needs of scenarios from economic to high-end.

[0004] In a first aspect, the present invention provides a control method for multi-sensor fusion environmental perception, comprising: Multimodal sensing data is collected through a sensor module pool integrated into a standard switch panel within the building space; the sensor module pool includes a radar sensor and a carbon dioxide sensor. Calculate the radar presence confidence level of the radar sensor and the carbon dioxide metabolism confidence level of the carbon dioxide sensor; An adaptive weighted fusion model is adopted, which dynamically allocates weights according to the currently configured sensor combination, and performs a weighted summation of the calculated confidence scores of the individual sensors to calculate the overall probability of the presence of a human body; Based on the relationship between the comprehensive probability of the human body's presence and a preset threshold, it is determined whether a human body exists in the target space, and the presence status of the human body is output. Based on the human body's state, adaptive control commands for environmental devices are generated and sent.

[0005] Furthermore, the overall probability of the human body's existence is calculated. The formula is: ; ; in, For the first Confidence level of available sensors, Dynamic weights; The overall environmental confidence level is calculated based on environmental sensor data. As weight; For historical confidence based on historical states, As weight; The dynamic weight Automatically assigned based on the currently active sensor combination mode, which includes a minimalist mode containing only the radar sensor, a dual-sensor mode containing both the radar sensor and the carbon dioxide sensor, and a three-sensor mode containing the radar sensor, the carbon dioxide sensor, and the temperature and humidity sensor.

[0006] Furthermore, the radar possesses confidence level. The calculation method includes: detecting the micro-motion characteristics of the target using millimeter-wave radar; when limb movements are detected, Assign a value of 1.0; when only respiratory micro-movements are detected, The value is assigned to 0.85; when only a slight heartbeat is detected, The value is assigned to 0.75; when no micro-motion features are detected, The value is assigned to 0.1; when a non-human interference source is detected, it is filtered through the spectrum feature library. Assign a value of 0.0; The confidence level of carbon dioxide metabolism The calculation method includes: calculating the rate of change of carbon dioxide concentration; when the rate of change is greater than a first threshold of 50 ppm / 10 minutes, The value is assigned to 0.8; when the rate of change is between the second threshold of 20 ppm / 10 minutes and the first threshold, The value is assigned to 0.5; when the rate of change is less than the second threshold, The value is assigned to 0.1; and the data from the temperature and humidity sensor is used to perform temperature compensation on the carbon dioxide concentration change rate to eliminate interference from carbon dioxide fluctuations caused by the fresh air from the air conditioner.

[0007] Furthermore, the method also includes a vital sign grading judgment step: based on the confidence level of the radar presence. The value is used to determine the level of vital signs: when When the value is 1.0, it is considered a high activity level, corresponding to a state of alertness and activity; when... When the value is 0.85, it is considered a low activity level, corresponding to a sedentary or sleep state; when... When the value is 0.75, it is judged as a slightly active level, corresponding to deep sleep or a resting state; when the human body has a comprehensive probability When the value is less than 0.30, it is determined to be at the level of no vital signs.

[0008] Furthermore, the method also includes a spatial region positioning step: Using at least three distributed switch panel nodes, the distance to the target human body is measured by the radar sensors within each node. , , ; The position coordinates (x, y) of the target human body in a preset two-dimensional coordinate system are calculated using the trilateration method; The building space is pre-divided into multiple functional areas. Based on the relationship between the location coordinates (x, y) and the boundaries of each functional area, the target human body is mapped to one or more corresponding functional areas. A fuzzy membership function is introduced. When the distance between the human body coordinates and the area boundary is less than 0.5 meters, a dual-area result of the main area and the secondary area is output to avoid boundary jumps.

[0009] Furthermore, the method also includes a sensor fault adaptive step: when any sensor in the sensor module pool is detected to have failed or has abnormal data, its corresponding dynamic weight is adjusted. The weights are reset to zero, and then redistributed among the remaining available sensors, environmental confidence, and historical confidence to maintain a total weight of 1, thus preserving the overall probability of the human body's presence. Robustness of computation.

[0010] Furthermore, the method also includes a self-learning, debug-free deployment step: After the switch panel is powered on, it enters a 72-hour baseline learning period, automatically collects and establishes an empty room baseline model, which includes background carbon dioxide concentration, radar static echo characteristics, temperature and humidity baselines, and light baselines. Automatically identify and record the characteristic spectrum of non-human interference sources, including air conditioner vents, curtains swaying, fans rotating, and pet activities. Incorporate the characteristics of the non-human interference sources into the filtering algorithm to reduce the false judgment rate. After each switch panel node is powered on, it automatically discovers neighboring nodes, forms a Mesh network, elects a master node, and completes network self-organization and self-repair.

[0011] Furthermore, the sensor modules in the sensor module pool adopt a unified standardized interface, which includes: a physical interface with a standardized slot, an electrical interface with a unified power supply voltage of 3.3V or 5V and a unified communication protocol of I2C, SPI or UART, and a software interface with a unified driver abstraction layer, supporting hardware hot-swapping or factory soldering configuration. The method also includes a software-defined sensor step: all sensor hardware interfaces and main control capabilities are pre-installed at the factory, and the functions of specific sensor modules are activated or unlocked as needed through software configuration or OTA remote upgrade, so as to realize the dynamic loading of sensor functions.

[0012] Furthermore, the adaptive control command is generated based on the human body's state, vital sign level, and spatial location results, including: When the location is determined to be occupied, a command is sent to turn on or adjust the lighting or air conditioning equipment in the corresponding area. When the area is determined to be unoccupied and the duration exceeds a preset threshold, a command is sent to turn off the lighting, air conditioning, or socket devices in the corresponding area. When the system is determined to be in sleep mode, it sends instructions to adjust the target temperature of the air conditioner, reduce the brightness of the lighting, or turn off the lighting. When a target human body is determined to have entered a preset privacy or danger zone, a security alarm or health monitoring command is sent.

[0013] Secondly, the present invention provides a multi-sensor fusion environmental perception control system, comprising: The perception layer includes one or more smart switch panel nodes distributed across various functional areas of the building space; Each panel node integrates a main control chip, a communication module, and a modular sensor module pool; the sensor module pool contains multiple optional sensor modules with a unified standardized interface, and the optional sensor modules include at least a millimeter-wave radar module and a carbon dioxide sensor module, and different sensor combinations are configured according to the needs of the scenario; The network layer connects each of the smart switch panel nodes with the decision layer. The network layer supports power line carrier communication or short-range wireless communication protocols and uses the original high-voltage lines of the switch panel to supply power and transmit data. The decision-making layer includes an edge computing gateway or a cloud computing server, which runs a multi-sensor fusion algorithm engine to execute the method steps and output the human body's presence status, vital sign level, and spatial location results. The execution layer connects to the controlled environment devices, receives adaptive control commands generated by the decision layer, and performs switching and adjustment operations on lighting devices, air conditioning devices, curtain devices, socket devices, or fresh air devices.

[0014] The technical effects of this invention are: 1. Eliminate ceiling wiring: By utilizing the existing 86-type junction box and power supply line on the switch panel, the renovation time for a single room is reduced from several hours to 20-30 minutes, and the renovation cost is reduced by more than 60%.

[0015] 2. Modular configuration: Hardware costs cover the entire price range, achieving "one platform, multiple configurations, and full scenario coverage".

[0016] 3. Multi-sensor fusion: By fusing data from multiple sources such as radar, carbon dioxide, temperature and humidity, the accuracy of vital sign judgment can reach over 98%, significantly reducing the false judgment rate.

[0017] 4. Distributed spatial positioning: Through multi-panel collaborative ranging and triangulation, it achieves zoned energy-saving control that "serves wherever people are," saving 15%-30% more energy than traditional whole-house switching.

[0018] 5. Self-learning and debugging-free: Built-in self-learning of environment baseline, self-learning of interference source feature library and network self-organizing mechanism, ordinary electricians can complete the installation and deployment. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating a multi-sensor fusion environmental perception control method proposed in an embodiment of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] The present invention aims to solve the technical problems in the prior art, such as the high cost of modification due to the need for independent wiring of sensors, the existence of detection blind spots and misjudgments by single sensors, the fixed and unadjustable sensor configuration, and the difficulty in achieving fine-grained control of zones.

[0022] refer to Figure 1 As shown, to achieve the above objectives, this invention provides a multi-sensor fusion environmental perception control method. In this embodiment, a hotel suite undergoes intelligent renovation. The suite includes an entrance hall, living room, master bedroom, bathroom, and work area. During the renovation, no ceiling openings or new wiring are made; only the existing ordinary mechanical switches are replaced with the intelligent switch panel of this invention. The panel is powered using the existing live wire (L) and neutral wire (N) in the 86 junction box.

[0023] Based on the hotel's functional requirements and budget, the following modular configuration was implemented: The main control panel in the entryway is configured for full-function mode. The main control chip inside this panel integrates the following sensor modules through standardized interfaces (physical interface: M.2 slot; electrical interface: 3.3V voltage and I2C protocol; software interface: unified driver abstraction layer): a 60GHz millimeter-wave radar module (planar microstrip array antenna, detection range 90° horizontally, 60° vertically, distance 8m), an NDIR carbon dioxide sensor module (range 400-5000ppm), a temperature and humidity sensor module (temperature accuracy ±0.5℃, humidity accuracy ±3%RH), a light sensor module (0.01-100,000lux), a noise sensor module (30-120dB), and a barometric pressure sensor module (accuracy ±0.06hPa).

[0024] Master bedroom control panel: Configured in standard mode. Integrates a 60GHz radar module, a carbon dioxide module, and a temperature and humidity module.

[0025] Bathroom panel: Configured in enhanced mode. It integrates a 24GHz waterproof radar module, a carbon dioxide module, and a pneumatic module.

[0026] Work area panel: Configured in basic mode. Only one 5.8GHz radar module is integrated. All panels form a sensing network via wireless or wired connections, requiring no additional communication cables. After the system powers on, each panel automatically discovers nearby nodes and elects the entrance panel as the master control node, completing the self-organizing network.

[0027] Among them, wired refers to PLC or microcontroller, while wireless includes Bluetooth, 2.4G or WIFI, etc.

[0028] The first 72 hours after the system is powered on is the baseline learning period, during which no manual intervention is required. Environmental baseline establishment: Each panel automatically collects background carbon dioxide concentration, static echo characteristics of radar in unoccupied conditions, temperature and humidity baseline values, and light baseline values ​​at its installation location. For example, the bathroom panel recorded an air pressure of 1013 hPa when unoccupied, and the bedroom panel recorded an ambient noise level of 30 dB at night. These data collectively constitute the "empty room baseline model" for that node.

[0029] Interference source learning: The system automatically identifies and records the spectral characteristics of non-human interference sources.

[0030] For example, if the radar module detects regular airflow disturbances from an air conditioner vent, and its spectral characteristics differ from human breathing, the system will label this as "air conditioner interference" and incorporate it into a filtering algorithm. When this characteristic is detected again subsequently, the radar will assess its confidence level. It was forcibly assigned the value 0.0.

[0031] Area mapping: Installers simply drag and drop on the mobile app to bind the MAC addresses of each panel to areas such as "entrance hall," "master bedroom," and "bathroom" on the suite floor plan. The system automatically establishes the spatial coordinate system.

[0032] The core decision-making algorithm of this invention runs in an edge computing gateway. The following uses a typical scenario of "a guest sleeping in the master bedroom" to illustrate the calculation and judgment process of the algorithm in detail.

[0033] Single sensor confidence calculation: Radar has confidence level The 60GHz radar module in the master bedroom transmits a frequency-modulated continuous wave and receives the echo. The algorithm performs a Fast Fourier Transform on the intermediate frequency signal, extracting a periodic micro-motion signal of 0.1-0.5Hz. This signal characteristic perfectly matches the rise and fall of the human chest cavity during breathing, without detecting any large-amplitude limb movements. According to the built-in rules, the system determines that "micro-breathing movement has been detected," therefore... The value is assigned to 0.85.

[0034] Confidence level of carbon dioxide metabolism The carbon dioxide sensor in the master bedroom samples once per minute. The system calculates the concentration change rate over the past 10 minutes. The initial concentration was 450 ppm (close to the outdoor background), and after 10 minutes it slowly rose to 520 ppm, with a change rate of (520−450) / 10 minutes = 7 ppm / minute. This value is greater than the set first threshold (5 ppm / minute, corresponding to 50 ppm / 10 minutes), and the temperature and humidity sensor data shows a slight temperature increase of 0.3℃, consistent with human body heat dissipation characteristics. Therefore, The value is assigned to 0.8.

[0035] Environmental overall confidence level The light sensor detected ambient light levels close to zero, with no abrupt changes; the noise sensor detected steady breathing sounds, with no abrupt changes. Therefore... The value is assigned to 0.

[0036] Based on the current sensor configuration of the master bedroom panel (three-sensor mode: radar + carbon dioxide + temperature and humidity), the system dynamically loads preset weights: =0.60, =0.30, =0.10, =0.0, =0.0. Historical confidence levels are ignored to simplify calculations.

[0037] The human body has a comprehensive probability Calculate using the following formula: ; Among them, the confidence level of the temperature and humidity sensor This factor serves as an auxiliary factor to verify the rationality of environmental changes; in this case, its value is 0.6. Substituting the values ​​into the calculation yields a final overall probability of human existence of 0.81.

[0038] because =0.81, which is greater than the preset "occupied" threshold of 0.70, so the system ultimately determines that the master bedroom is "occupied".

[0039] At the same time, based on A value of 0.85 indicates that, according to the vital signs grading rules, the system determines the current state to be Level 2 (low activity), which corresponds to "sleep state".

[0040] When a guest gets up from the master bedroom and walks towards the bathroom, the system performs spatial area positioning: Single-panel ranging: The radar modules on the master bedroom panel, living room panel, and bathroom panel simultaneously detect moving targets. Each panel calculates the distance to the guest based on the radar echo time delay and signal strength (RSSI). , , .

[0041] Triangulation: The system acquires three distance values ​​and uses trilateration to calculate the guest's precise position coordinates (x, y) in a preset two-dimensional coordinate system. The core of the algorithm is to solve for the intersection of three circles, as shown in the following system of equations: ; in, , , These are the known coordinates of the three panels.

[0042] Area Mapping and Control: The system calculates that the coordinates (x, y) are located within the "toilet" area. The decision-making layer then generates control commands: Send a command to the execution layer of the "master bedroom": turn off the main lighting in the master bedroom and adjust the air conditioning temperature from 22℃ to 24℃ (enter unattended energy-saving mode).

[0043] Send a command to the execution layer of the "bathroom": turn on the 3000K color temperature night light, adjust the brightness to 30%, and turn on the exhaust fan.

[0044] It achieves a refined zoning control effect of "lights off when people leave and preheating before people arrive".

[0045] Fault Adaptation: Assume that during operation, the carbon dioxide sensor on the bathroom panel experiences constant data due to dust accumulation. The system monitoring module detects that the sensor's data variance is 0 within one hour, classifying it as a fault. The system automatically adjusts the sensor's dynamic weight... The weight is reset to zero, and the released 0.30 weight is proportionally redistributed to the remaining radar and air pressure modules in the bathroom panel, so that the total weight remains 1. The determination of human presence in the bathroom continues to be supported by radar and air pressure data (e.g., determining door opening / closing through air pressure fluctuations), and the system maintains overall robustness.

[0046] Software-defined upgrade: In the early stages of hotel operation, due to cost considerations, only the basic radar mode was activated on the work area panel. After three months of operation, it was found that employees frequently spent long periods sitting and working in the work area, and the single-radar mode posed a risk of accidentally turning off lights. Without replacing any hardware, the hotel remotely activated the pre-installed carbon dioxide sensor function on the work area panel via OTA (Over-The-Air) upgrade after paying the software license fee. The panel's sensor combination was switched online from "minimalist mode" to "dual-sensor mode," with improved weighting. and Automatic updates immediately improve the accuracy of human presence detection.

[0047] Through the detailed description of the above specific embodiments, those skilled in the art can clearly understand that the present invention, by integrating modular multi-sensors into a switch panel and combining self-learning deployment, adaptive weighted fusion algorithm, spatial triangulation positioning and software-defined upgrade and other technical means, systematically solves the problems of high transformation cost, detection misjudgment, coarse control and rigid configuration in the prior art, and achieves the intended technical effect.

[0048] Based on the same inventive concept, this invention provides a multi-sensor fusion environmental perception control system, comprising: The perception layer includes one or more smart switch panel nodes distributed across various functional areas of the building space; Each panel node integrates a main control chip, a communication module, and a modular sensor module pool; the sensor module pool contains multiple optional sensor modules with a unified standardized interface, and the optional sensor modules include at least a millimeter-wave radar module and a carbon dioxide sensor module, and different sensor combinations are configured according to the needs of the scenario; The network layer connects each of the smart switch panel nodes with the decision layer. The network layer supports power line carrier communication or short-range wireless communication protocols and uses the original high-voltage lines of the switch panel to supply power and transmit data. The decision-making layer includes an edge computing gateway or a cloud computing server, which runs a multi-sensor fusion algorithm engine to execute the method steps and output the human body's presence status, vital sign level, and spatial location results. The execution layer connects to the controlled environment devices, receives adaptive control commands generated by the decision layer, and performs switching and adjustment operations on lighting devices, air conditioning devices, curtain devices, socket devices, or fresh air devices.

[0049] This is the first time that the two technical features of "modular multi-sensor configuration" and "standard switch panel integration" have been combined. By reserving a unified and standardized interface inside the standard 86 switch panel, users or engineers can freely choose configurations from single sensors to full-featured multi-sensors according to their budget and scenario requirements.

[0050] Example embodiments have been disclosed herein, and while specific terminology has been used, it is for illustrative purposes only and should be construed as such, and is not intended to be limiting. In some instances, it will be apparent to those skilled in the art that features, characteristics, and / or elements described in conjunction with particular embodiments may be used alone, or in combination with features, characteristics, and / or elements described in conjunction with other embodiments, unless otherwise expressly indicated. Therefore, those skilled in the art will understand that various changes in form and detail may be made without departing from the scope of the invention as set forth in the appended claims.

Claims

1. A control method for multi-sensor fusion environmental perception, characterized in that, include: Multimodal sensing data is collected through a pool of sensor modules integrated into standard switch panels within the building space; The sensor module pool includes at least a radar sensor and a carbon dioxide sensor; Calculate the radar presence confidence level of the radar sensor and the carbon dioxide metabolism confidence level of the carbon dioxide sensor; An adaptive weighted fusion model is adopted, which dynamically allocates weights according to the currently configured sensor combination, and performs a weighted summation of the calculated confidence scores of the individual sensors to calculate the overall probability of the presence of a human body; Based on the relationship between the comprehensive probability of the human body's presence and a preset threshold, it is determined whether a human body exists in the target space, and the presence status of the human body is output. Based on the state of the human body, adaptive control commands for environmental devices are generated and sent.

2. The method according to claim 1, characterized in that, Calculate the overall probability of the human body's existence. The formula is: ； ； in, For the first Confidence level of available sensors, Dynamic weights; The overall environmental confidence level is calculated based on environmental sensor data. As weight; For historical confidence based on historical states, As weight; The dynamic weight Automatically assigned based on the currently active sensor combination mode, which includes a minimalist mode containing only the radar sensor, a dual-sensor mode containing both the radar sensor and the carbon dioxide sensor, and a three-sensor mode containing the radar sensor, the carbon dioxide sensor, and the temperature and humidity sensor.

3. The method according to claim 1, characterized in that, The radar has confidence level. The calculation method includes: detecting the micro-motion characteristics of the target using millimeter-wave radar; when limb movements are detected, Assign a value of 1.0; when only respiratory micro-movements are detected, The value is assigned to 0.85; when only a slight heartbeat is detected, The value is assigned to 0.75; when no micro-motion features are detected, The value is assigned to 0.1; when a non-human interference source is detected, it is filtered through the spectrum feature library. Assign a value of 0.0; The confidence level of carbon dioxide metabolism The calculation method includes: calculating the rate of change of carbon dioxide concentration; when the rate of change is greater than a first threshold of 50 ppm / 10 minutes, The value is assigned to 0.8; when the rate of change is between the second threshold of 20 ppm / 10 minutes and the first threshold, The value is assigned to 0.5; when the rate of change is less than the second threshold, The value is assigned to 0.1; and the data from the temperature and humidity sensor is used to perform temperature compensation on the carbon dioxide concentration change rate to eliminate interference from carbon dioxide fluctuations caused by the fresh air from the air conditioner.

4. The method according to claim 1, characterized in that, The method also includes a vital signs grading judgment step: based on the confidence level of the radar presence. The value is used to determine the level of vital signs: when When the value is 1.0, it is considered a high activity level, corresponding to a state of alertness and activity; when... When the value is 0.85, it is considered a low activity level, corresponding to a sedentary or sleep state; when... When the value is 0.75, it is judged as a slightly active level, corresponding to deep sleep or a resting state; when the human body has a comprehensive probability When the value is less than 0.30, it is determined to be at the level of no vital signs.

5. The method according to claim 1, characterized in that, The method also includes a spatial region positioning step: Using at least three distributed switch panel nodes, the distance to the target human body is measured by the radar sensors within each node. , , ; The position coordinates (x, y) of the target human body in a preset two-dimensional coordinate system are calculated using the trilateration method; The building space is pre-divided into multiple functional areas. Based on the relationship between the location coordinates (x, y) and the boundaries of each functional area, the target human body is mapped to one or more corresponding functional areas. A fuzzy membership function is introduced. When the distance between the human body coordinates and the area boundary is less than 0.5 meters, a dual-area result of the main area and the secondary area is output to avoid boundary jumps.

6. The method according to claim 1, characterized in that, The method further includes a sensor fault adaptation step: when any sensor in the sensor module pool is detected to have a fault or data anomaly, its corresponding dynamic weight is adjusted. The weights are reset to zero, and then redistributed among the remaining available sensors, environmental confidence, and historical confidence to maintain a total weight of 1, thus preserving the overall probability of the human body's presence. Robustness of computation.

7. The method according to claim 1, characterized in that, The method also includes a self-learning, debug-free deployment step: After the switch panel is powered on, it enters a 72-hour baseline learning period, automatically collects and establishes an empty room baseline model, which includes background carbon dioxide concentration, radar static echo characteristics, temperature and humidity baselines, and light baselines. Automatically identify and record the characteristic spectrum of non-human interference sources, including air conditioner vents, curtains swaying, fans rotating, and pet activities. Incorporate the characteristics of the non-human interference sources into the filtering algorithm to reduce the false judgment rate. After each switch panel node is powered on, it automatically discovers neighboring nodes, forms a Mesh network, elects a master node, and completes network self-organization and self-repair.

8. The method according to claim 1, characterized in that, The sensor modules in the sensor module pool adopt a unified standardized interface, which includes: a physical interface with a standardized slot, an electrical interface with a unified power supply voltage of 3.3V or 5V and a unified communication protocol of I2C, SPI or UART, and a software interface with a unified driver abstraction layer, supporting hardware hot-swapping or factory soldering configuration. The method also includes a software-defined sensor step: all sensor hardware interfaces and main control capabilities are pre-installed at the factory, and the functions of specific sensor modules are activated or unlocked as needed through software configuration or OTA remote upgrade, so as to realize the dynamic loading of sensor functions.

9. The method according to claim 1, characterized in that, The adaptive control command is generated based on the human body's physical state, vital sign level, and spatial location results, and includes: When the location is determined to be occupied, a command is sent to turn on or adjust the lighting or air conditioning equipment in the corresponding area. When the area is determined to be unoccupied and the duration exceeds a preset threshold, a command is sent to turn off the lighting, air conditioning, or socket devices in the corresponding area. When the system is determined to be in sleep mode, it sends instructions to adjust the target temperature of the air conditioner, reduce the brightness of the lighting, or turn off the lighting. When a target human body is determined to have entered a preset privacy or danger zone, a security alarm or health monitoring command is sent.

10. A control system for multi-sensor fusion environmental perception, characterized in that, include: The perception layer includes one or more smart switch panel nodes distributed across various functional areas of the building space; Each panel node integrates a main control chip, a communication module, and a modular sensor module pool; the sensor module pool contains multiple optional sensor modules with a unified standardized interface, and the optional sensor modules include at least a millimeter-wave radar module and a carbon dioxide sensor module, and different sensor combinations are configured according to the needs of the scenario; The network layer connects each of the smart switch panel nodes with the decision layer. The network layer supports power line carrier communication or short-range wireless communication protocols and uses the original high-voltage lines of the switch panel to supply power and transmit data. The decision layer includes an edge computing gateway or a cloud computing server, which runs a multi-sensor fusion algorithm engine to execute the method of any one of claims 1 to 9 and output the human body presence status, vital sign level and spatial area positioning results. The execution layer connects to the controlled environment devices, receives adaptive control commands generated by the decision layer, and performs switching and adjustment operations on lighting devices, air conditioning devices, curtain devices, socket devices, or fresh air devices.

Images