Guest room electricity taking method based on radar identification of number and trend of space personnel

Abstract

The invention discloses a guest room electricity taking method based on radar identification of the number and trend of space personnel, and relates to the technical field of intelligent hotel control. According to the method, a guest room layout is obtained, a key area is divided, and an intelligent switch containing a radar sensor and an induction area are configured; a radar sensor collects a space detection signal, and the existence state, the real-time number and the moving direction of personnel are analyzed through processing; and when the sensing area meets the triggering condition, activating a corresponding equipment linkage list, generating an electricity taking control instruction and a scene mode switching instruction, controlling the on-off of a power supply system, and linking the intelligent equipment to switch an operation mode. Through high-precision radar sensing, multi-dimensional data processing and intelligent scene linkage, full-process automation of'human sensing-analysis-control 'is achieved, the problems that a traditional system is insufficient in personnel recognition precision, lack of multi-target analysis and low in linkage efficiency are solved, and guest room energy efficiency and service quality are improved.

Description

Technical Field

[0001] This application relates to the field of intelligent hotel room control technology, and more specifically, to a method and system for powering a guest room based on radar identification of the number and movement of people in the space. Background Technology

[0002] With the rapid advancement of technology and the increasing demands for quality travel experiences, intelligent management of hotel rooms has become an inevitable trend in the industry. Traditional hotel room equipment control models have significant shortcomings. Each device operates independently, lacking scenario-based linkage and data exchange, resulting in low management efficiency, serious energy waste, and difficulty in meeting the needs of refined operations.

[0003] Current mainstream solutions primarily rely on infrared sensors or single radar detection technology. These solutions can only achieve simple occupancy / absence detection and cannot accurately identify vital signs of a person in a stationary state. When guests are asleep or sitting still, the system is prone to misjudging the situation, causing equipment to shut down abnormally, severely impacting the user experience. More importantly, existing technologies lack the ability to identify the spatial distribution of multiple targets and cannot distinguish the activity characteristics of people in different areas of the guest room, making it impossible to achieve differentiated control for different functional areas such as sleeping areas and bathroom areas. In addition, existing solutions using infrared sensors or single radar detection technology require additional wiring based on the number of infrared sensors or single radars, increasing the initial wiring process and construction time, which is detrimental to improving installation efficiency and cost savings.

[0004] Chinese patent publication CN111522002A discloses an intelligent switch radar system. This system uses millimeter-wave radar sensors to transmit and receive radar waves to detect the presence of a human body. The main control circuit board of the intelligent switch controls the controlled electrical appliances based on the detection results, eliminating the need for user pre-setting or sending commands. This achieves a certain degree of automated control and represents a breakthrough in convenience and energy efficiency. Its core lies in using radar sensing technology to replace traditional manual triggering methods, providing a new approach to intelligent control of guest room equipment. However, this technology still has significant limitations in its deep application in hotel guest room scenarios: First, the perception dimension is limited, only able to make a binary judgment of "occupied / unoccupied," and cannot recognize the vital signs (such as breathing and slight movements) of stationary human bodies (such as those in a sleeping or sitting state). This can easily lead to situations where "a person is not present when they are stationary," causing the equipment to shut down erroneously and affecting the user experience. Second, it lacks multi-target analysis capabilities, making it impossible to accurately count the number of people in the guest room in real time, let alone distinguish the movement trajectories and behavioral characteristics of different individuals. This makes it difficult to meet the needs of zoning and environmental adjustment when multiple people are staying (such as differentiated control of children's areas and adult areas). Third, its system linkage logic is simple, based only on basic signal control equipment that controls whether something is "present or not". It does not form a deep connection with the layout of key areas in the guest room (such as the sleeping area and activity area) and the dynamic behavior of people (such as the length of stay and the direction of movement), and cannot realize the automatic switching of scene modes such as welcoming and sleeping. Therefore, overcoming the shortcomings of existing radar sensing technology in terms of personnel identification accuracy, multi-target analysis, and scene linkage depth has become an important issue that urgently needs to be addressed in the field of intelligent hotel room management. Summary of the Invention

[0005] To address the aforementioned problems, this invention proposes a method for powering guest rooms based on radar-based identification of the number and movement of people in the space. Through radar sensing, multi-dimensional data processing, and intelligent scene linkage, it achieves full-process automation of "human perception-analysis-control," aiming to improve the energy efficiency and service quality of hotel guest rooms.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: This invention discloses a method for controlling power supply in guest rooms based on radar identification of the number and movement of people in the space, including: Step S1: Obtain the layout plan of the target guest room, identify key areas of the guest room, and configure one or more smart switches according to the key areas of the guest room, while dividing the sensing areas; wherein, the smart switch includes at least: a radar sensor, a central control module, and a network module, and each smart switch in the sensing area is configured with a corresponding device linkage list; Step S2: The smart switch collects spatial detection signals in the guest room in real time through radar sensors and generates real-time sensing data. It also processes the spatial detection signals and real-time sensing data to obtain information on the presence status of people in the guest room, the real-time number of people, and the direction of movement. The spatial detection signals include human vital signs signals, movement trajectory signals, and multi-target distribution signals. Step S3: When the real-time sensing data of a certain sensing area meets the preset trigger conditions, activate the device linkage list corresponding to that area. Step S4: Generate a power supply control command based on the personnel presence status, real-time number of people, and movement direction information, and generate a scene mode switching command based on the activated device linkage list; Step S5: Control the on / off state of the guest room power supply system based on the power supply control command, and switch the smart electrical equipment in the guest room to the corresponding operating mode according to the scene mode switching command.

[0007] Preferably, the key areas of the guest room mentioned in step S1 include: the entrance area, the sleeping area, the activity area, and the bathroom area; the sensing area is the detection range formed according to the scanning angle of each smart switch.

[0008] Preferably, step S2 includes: S21. The radar sensor emits electromagnetic waves and receives reflected signals to obtain the raw ADC signal as the basic data for space detection, covering the initial signals related to human vital signs, movement trajectories and multi-target distribution in the guest room. S22. Perform CFAR constant false alarm rate detection on the original ADC signal and filter out environmental noise and clutter interference through adaptive threshold to select effective signals related to human activities. S23. Perform a fast Fourier transform on the filtered signal to convert the time domain signal into frequency domain data, analyze the spatial location information including the distance and angle of the target and generate spatial point cloud data to construct a spatial location framework of the human target in the guest room. S24. Based on the generated spatial point cloud data, the DBSCAN point cloud clustering algorithm is used to cluster the discrete target point clouds in space into independent individuals to achieve multi-target differentiation, thereby counting the real-time number of people in the guest rooms to obtain the number of people statistics information. S25. While S24 is being executed, Doppler phase analysis is performed based on the generated spatial point cloud data. Phase difference information is extracted using the Doppler effect, the human body's movement direction vector is calculated, and the human body's movement trajectory and direction are analyzed. S26. Integrate spatial location information, people statistics, and movement direction vectors to form real-time sensing data that includes people's presence status, real-time number of people, and movement direction information.

[0009] Preferably, step S2 further includes: S27. Based on the real-time sensing data, perform in-depth analysis of pedestrian flow direction and behavior prediction, specifically including: S271. Based on the spatial position information and movement direction vector in the real-time sensing data, assign a spatial position weight to the detected target at a certain position according to the spatial position information and a preset weight value, and further analyze the movement speed of the detected target according to the movement direction vector and generate the target velocity vector by combining the movement direction vector. S272. Perform dynamic entropy analysis and calculate the movement trend value: The movement trend value is calculated as the sum of the products of the target velocity vector and the spatial position weight. S273. Construct a behavior prediction model and obtain the trajectory entropy value by calculating the rate of change of the movement trend value over n consecutive time frames. When the trajectory entropy value is greater than the preset threshold, it is judged as a highly random movement and an emergency evacuation tag is triggered. When the trajectory entropy value is less than or equal to a preset threshold, it is determined to be a regular activity and an energy-saving scenario label is generated. The trajectory entropy analysis results and output labels are integrated into the real-time sensing data.

[0010] Preferably, step S3 includes: S31. Receive the real-time sensing data output in step S2 in real time, and extract human activity feature parameters in each sensing area. The human activity feature parameters include human movement speed, dwell time, target quantity change rate and activity area coverage. S32. Compare and analyze the extracted human activity feature parameters with the pre-set trigger conditions for the sensing area: When the human body's movement speed is greater than the first speed preset threshold and the duration is greater than or equal to the first time preset threshold, it is determined to be "dynamic activity triggered"; When the human body's movement speed is less than or equal to the first speed preset threshold and the duration is greater than or equal to the second time preset threshold, it is determined to be "static dwell trigger". When the number of targets in the sensing area increases from zero, it is determined as "new entry triggered"; When the number of targets in all sensing areas decreases from present to absent, it is determined as "departure trigger".

[0011] S33. If any of the above triggering conditions are met, the central control module calls the pre-configured device linkage list of the sensing area through the network module. The device linkage list includes the trigger priority order of each smart electrical device and the relevant parameter threshold settings. S34. For sensing areas that do not meet the trigger conditions, maintain the current operating state of the device and continuously monitor changes in its characteristic parameters. When no human activity characteristics are detected for a continuous period of time, mark it as a low-activity area.

[0012] Preferably, the "power control command" in step S4 includes the following generation modes: Continuous power supply mode: In response to "dynamic activity trigger" and "new entry trigger" states, a continuous power supply command is generated to keep the guest room power supply system on. Energy-saving power supply mode: In response to the "static dwell trigger" state, an energy-saving power supply command is generated, which automatically reduces the power supply of smart electrical devices outside the sensing area; Delayed power-off mode: In response to the "away trigger" state, a delayed power-off command is generated, and after a preset delay, non-essential power supply is cut off, while power supply to critical equipment is preserved.

[0013] Preferably, the "scene mode switching instruction" in step S4 is a specific execution instruction for the device linkage list activated in step S3, including preset modes based on the activity status of people in different areas: Welcome Mode: This mode controls the lighting, curtains, and air conditioning equipment to switch to the state appropriate for when a person enters the area, based on the "new entry trigger" device list at the entrance. Sleep Mode: A list of devices linked to the "static dwell trigger" in the corresponding sleep area to control lighting, curtains, and air conditioning to switch to a state suitable for rest; Activity Mode: The device linkage list corresponding to the "Dynamic Activity Trigger" in the activity area, to control the lighting and entertainment equipment to switch to the operating state that is suitable for multi-person interaction or activities; Bathroom Mode: This mode controls the bathroom lighting and ventilation equipment to switch to a state suitable for washing and bathing scenarios, based on the list of devices triggered by "new entry" or "static stay" in the bathroom area. Energy Saving Mode: For the device linkage list marked "Low Activity Area", control the power of non-essential electrical equipment in the area to reduce or turn off to match the low activity demand.

[0014] Leave Mode: The list of devices linked to the "Leave Trigger" in the corresponding doorway area, controlling the devices to switch to the state where the personnel are leaving.

[0015] Preferably, the radar sensor is a 60GHz human presence sensing radar sensor.

[0016] Preferably, the radar sensor includes: a Fresnel lens beamforming unit 11, a MIMO antenna array 12, and an algorithm processing unit 13; Among them, the Fresnel lens beamforming unit 11 enhances the directional reception capability of spatial presence sensing signals, thereby enhancing the micro-motion signals generated by human life activities, so as to realize the recognition of the static / active state of the human body. The MIMO antenna array 12 is used to acquire multi-dimensional human movement trajectory signals and calculate the human body azimuth angle by the phase difference of the multi-channel received signals to obtain the human body's azimuth information relative to the radar. The algorithm processing unit 13 loads a multi-target point cloud separation algorithm and a neural network model. The multi-target point cloud separation algorithm performs density clustering on the spatial point cloud data collected by the MIMO antenna array 12, and distinguishes different human targets by setting a distance threshold to complete the number of people count. The neural network model receives continuous azimuth angle data, constructs a human flow line model, and outputs a human activity heat map and trajectory prediction results to provide a basis for human movement for power supply control.

[0017] Preferably, step S27, the in-depth analysis of pedestrian flow and behavior prediction, further includes the following steps: Real-time sensing data is uploaded to the SaaS system via a network module. The SaaS system then calls a preset big data analysis model to jointly analyze the historical sensing data and real-time data, optimizing the trajectory entropy calculation parameters and preset thresholds in the behavior prediction model. The big data analysis model dynamically adjusts the judgment logic under different scenarios based on the activity feature samples of multiple guest rooms, so that the accuracy of distinguishing between highly random movement and regular activities is adapted to the actual use scenario of the guest rooms.

[0018] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: 1. Addressing the issues of existing technologies requiring additional wiring, complex installation, and poor compatibility, this solution employs an integrated radar sensor smart switch design. Installation is completed without additional wiring, reducing upfront construction costs and time, simplifying installation and maintenance, and eliminating the need to alter existing room layouts, thus improving guest room adaptability. Simultaneously, the system supports multi-brand smart device integration. Through a network module and interface with a SaaS system, it can integrate multi-guest room data for global optimization, adapting to different guest room types and usage scenarios. In case of device failure, it can quickly locate and push maintenance information, reducing maintenance costs and significantly lowering the barrier to entry for hotel smart technology upgrades.

[0019] 2. Addressing the limitations of traditional infrared sensors or single radars in the background technology, which can only determine "occupancy / vacancy" and cannot identify the vital signs of stationary human beings or multiple targets, this solution utilizes radar sensors combined with multi-dimensional data processing technology to achieve accurate perception of the status of guests. It can capture signals of stationary states such as breathing and subtle movements, effectively distinguishing between sleeping and sitting scenarios, avoiding misjudgments such as "power outage due to occupancy" or "power consumption by no one." Simultaneously, it uses point cloud clustering algorithms to differentiate multiple targets, accurately count the number of people in real time, and track individual movement trajectories, solving the pain point of being unable to identify different areas when multiple people are staying, and providing a reliable basis for refined control.

[0020] 3. This solution addresses the shortcomings of existing technologies, such as simplistic linkage logic and lack of integration with regional layout and personnel dynamics. It divides sensing ranges based on key areas of the guest rooms and constructs a multi-scenario automatic switching mechanism by combining trigger conditions such as "dynamic activity" and "static stay." For example, the entrance area automatically activates a welcome mode when a new person enters; the sleeping area switches to sleep mode when static stay is detected; and the bathroom area activates lighting and ventilation equipment based on the activity status of guests. This deep integration of "area-behavior-device" enables precise device linkage without manual operation, improving the user check-in experience and reducing the hotel's manual management costs. This achieves scenario-based intelligent linkage, enhancing both user experience and management efficiency.

[0021] 4. This solution addresses the issue of high energy consumption due to independent equipment operation in traditional models. By monitoring the activity status of people in each area in real time, it achieves refined energy consumption management. For spaces marked as "low-activity areas," it automatically reduces the power of unnecessary equipment or shuts it off; when people leave, it activates a delayed power-off mode, preserving power to critical equipment while cutting off unnecessary energy consumption; and it adjusts equipment operation strategies based on the activity status of multiple people to avoid indiscriminate energy consumption throughout the house. This dynamic control mechanism effectively reduces ineffective energy consumption in empty rooms and energy waste in inactive areas, balancing user experience and energy-saving needs. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a schematic diagram of the method flow of the present invention.

[0024] Figure 2 This is a schematic diagram of the structure after the method of the present invention is applied to the guest room power supply system. Detailed Implementation

[0025] 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.

[0026] It should be noted that the terms "first," "second," "third," "fourth," etc., used in the specification and claims of this invention are used to distinguish different objects, not to describe a specific order. The terms "comprising" and "having," and any variations thereof, in the embodiments of this invention are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products, or devices.

[0027] In existing technologies, intelligent management of hotel rooms is becoming increasingly common. However, traditional equipment control modes suffer from problems such as independent equipment operation and insufficient scene linkage. Existing radar sensing technology can only determine whether a room is occupied or unoccupied, but it cannot identify the vital signs of a stationary person, leading to frequent misjudgments. For example, a sleeping person may be mistakenly identified as unoccupied, causing equipment to shut down unnecessarily. Furthermore, existing systems lack multi-target analysis capabilities, failing to count real-time occupancy or differentiate individual behaviors, making it difficult to meet the zoning adjustment needs when multiple people are staying. The linkage logic is simple, relying solely on the presence of signals to control equipment, and cannot combine area layout and personnel dynamics to achieve scene switching.

[0028] To address these issues, it is necessary to enhance the radar's perception capabilities, solve the challenge of identifying stationary human figures, and simultaneously achieve multi-target tracking and behavior analysis.

[0029] like Figure 1 As shown, this case provides a method for power supply in hotel rooms based on radar-based identification of the number and movement of people in the space, including: Step S1: Obtain the CAD layout plan of the target guest room, identify key areas of the guest room, and configure one or more smart switches according to these key areas, while simultaneously dividing the sensing areas. Key areas refer to the entrance, sleeping, activity, and bathroom areas, which are defined according to the guest room's functions. Identifying these areas through the floor plan allows for targeted configuration of device linkage strategies. The sensing area refers to the detection range formed by the radar scanning angle of the smart switch; different areas are used to achieve zoned monitoring. In specific implementation, obtain the CAD layout plan of the target guest room, identify key areas such as the entrance, sleeping (bedside area), activity (living room), and bathroom; deploy at least one smart switch on each of the following: one side next to the entrance, the left and right walls of the sleeping area, and the TV background wall; and divide the sensing areas according to the radar scanning angle (120° horizontally, 90° vertically). Furthermore, such as Figure 2 As shown, the smart switch includes at least: a radar sensor 1, a central control module 2, and a network module 3. The radar sensor is responsible for signal acquisition, the central control module processes the data, and the network module enables device linkage. Each smart switch in the sensing area is configured with a corresponding device linkage list. The device linkage list contains the associated control logic for controlling the operation of smart electrical devices, such as associating corridor lights and curtains in the doorway area, and bedside lamps and air conditioners in the sleeping area, to match different scenario needs.

[0030] Step S2: The smart switch collects spatial detection signals in the guest room in real time through radar sensors and generates real-time sensing data. It also processes the spatial detection signals and real-time sensing data to obtain information on the presence status of people in the guest room, the real-time number of people, and their direction of movement. The spatial detection signals include human vital signs, movement trajectory signals, and multi-target distribution signals. Multi-dimensional signal fusion improves recognition accuracy. Specifically, the radar sensor emits electromagnetic waves 20 times per second and generates a raw ADC signal after receiving reflected signals. CFAR detection filters out furniture reflection noise, and Fourier transform is used to analyze the distance between the human body and the radar (e.g., 2.5m from the radar in the sleeping area) and the angle (30°), generating spatial point cloud data. The point cloud is clustered using the DBSCAN algorithm (Euclidean distance threshold 0.8m), and the number of people is counted as 2. The movement direction vector (e.g., the trajectory and direction of a target person moving towards the bathroom area) is calculated using Doppler phase difference, and the data is integrated to form real-time sensing data.

[0031] Step S3: When the real-time sensing data of a certain sensing area meets the preset trigger conditions (such as personnel movement speed > 0.3m / s, dwell time > 5s), activate the corresponding device linkage list for that area, such as adjusting the lighting brightness to 80% or deactivating the TV from standby mode. Step S4: Generate a power supply control command based on the personnel presence status, real-time number of people, and movement direction information, and generate a scene mode switching command based on the activated device linkage list; wherein, the scene mode switching command is generated according to the priority order of the linkage list, prioritizing the adjustment of the lighting brightness of the corresponding area, and then linking the adjustment of the air conditioning temperature of the associated area; for example, based on the sensing data of "person present, 2 people, moving towards the bathroom area", a "continuous power supply command" is generated; and an "entertainment mode switching command" is generated based on the activated device linkage list.

[0032] Step S5: Control the on / off state of the guest room power supply system based on the power control command, and link the electrical equipment in the guest room to switch to the corresponding operating mode according to the scene mode switching command. In specific implementation, the layout plan divides key areas and configures smart switches to form a monitoring network covering the entire guest room. Radar sensors continuously collect spatial signals, and after signal processing, analyze the status, number, and movement direction of personnel. When personnel activity that meets preset conditions is detected in a certain area, such as entering behavior detected at the door, the central control module calls the corresponding equipment linkage list and controls the lighting, air conditioning, and other equipment to start. At the same time, the power supply strategy is adjusted according to the real-time number of people, for example, maintaining full power supply when there are many people and delaying the shutdown of unnecessary equipment when there is no one. By integrating spatial location, number of people statistics, and movement direction data, power control commands and scene modes are dynamically generated to achieve precise matching between equipment control and personnel behavior.

[0033] As mentioned above, this solution first installs radar sensors on existing smart switches capable of controlling intelligent electrical devices, eliminating the need for additional wiring and reducing initial installation costs. The radar-equipped smart switches allow for automatic power supply and scene switching based on real-time occupant status, reducing user operating costs, as no manual card insertion or switch operation is required. Furthermore, by dividing the sensing range into key areas of the guest room and configuring dedicated device linkage lists, indiscriminate power supply throughout the room is avoided. For example, when only active areas are occupied, devices in the sleeping area maintain low power consumption, reducing unnecessary energy consumption. Simultaneously, the collaborative operation of multiple smart switches in this solution ensures comprehensive coverage of occupant movement paths, preventing issues such as lights not turning on when someone is present or devices continuing to operate when someone leaves, thus improving user experience and energy efficiency. In this way, the solution constitutes a closed-loop design encompassing "area configuration - radar sensing - data processing - linkage control," automating guest room power supply and device control.

[0034] As a specific implementation method, the key areas of the guest room mentioned in step S1 are specifically divided into: Entrance area: A fan-shaped area with a radius of 1.5m centered on the door, serving as a passageway for people to enter and exit, and connected to corridor lights, smart door locks, and electric curtains; Sleeping area: A 1.2m x 2m rectangular area on both sides of the headboard, at least covering the area where the bed is located, and including bedside lamps and air conditioning; Activity area: The area around the sofa and coffee table in the living room, including the main light, TV, and curtains; Bathroom area: The entire bathroom area, including the washroom space, and related features such as vanity lights and exhaust fans.

[0035] The sensing area is the detection range formed by the scanning angle of each smart switch, determined by the radar scanning angle. For example, with a horizontal scanning angle of 100° for the doorway radar, the smart switch covers the area from the entrance hall to the doorway; with a vertical scanning angle of 80° for the sleep area radar, switch B covers the area from the head of the bed to the window. This ensures no blind spots. Specific divisions must be combined with technical parameters such as the radar scanning angle (e.g., 90°-120°) and detection distance (e.g., 0.5-12 meters) to ensure that each area can be effectively monitored by the radar.

[0036] As mentioned above, compared with existing technologies, traditional solutions only monitor guest rooms as a single space, failing to distinguish between different behavioral scenarios such as entering and exiting the door, sleeping and silent, and active communication. This solution, by dividing key areas and configuring directional scanning sensing areas, enables radar sensors to optimize signal acquisition parameters for different functional zones. Specifically, by binding key areas of the guest room to a device linkage list, it ensures that device control only applies to the area where the target person is located, avoiding ineffective operation of devices in irrelevant areas and improving energy efficiency. At the same time, the sensing areas are divided according to the radar scanning angle to avoid signal interference between different areas, improving the positioning accuracy of the target person and providing reliable spatial data for subsequent scene switching. Moreover, by installing the aforementioned smart switches in each key area of ​​the guest room, it is easy to cover all key areas of the guest room in all scenarios, ensuring seamless perception throughout the entire process "from entering the room to leaving the room," and improving the continuity of the user experience.

[0037] Through the above technical solution, this application solves the problem of false triggering of controls caused by overlapping area functions in traditional guest room monitoring. Directional scanning in the entrance area can accurately distinguish between personnel entering / exiting and indoor movement, avoiding false triggering of the leave mode; high-sensitivity monitoring in the sleep area can identify vital signs in a resting state, preventing accidental shutdown of air conditioning or lighting equipment; multi-target tracking in the activity area can count the number of people in real time, providing a basis for adjusting the power of entertainment equipment; vertical scanning in the bathroom area can distinguish between standing and showering behavior, enabling precise start / stop of ventilation equipment. The range of each sensing area is limited based on the scanning angle, avoiding signal interference between different areas, allowing the device linkage list to execute differentiated control strategies according to spatial location.

[0038] like Figure 1 As shown, step S2 includes: S21. The radar sensor emits electromagnetic waves and receives reflected signals to acquire raw ADC signals, which serve as the basic data for spatial detection, covering initial signals related to human vital signs, movement trajectories, and the distribution of multiple targets within the guest room. The raw ADC signal refers to the unprocessed analog signal after the radar sensor receives the reflected waves. Specifically, it can be generated by sampling and quantizing the electromagnetic wave reflected signals using an analog-to-digital converter (ADC) to carry the raw information of human vital signs and movement trajectories. For example, the radar sensor emits 60GHz electromagnetic waves, receives the reflected signals, and converts them into 16-bit raw ADC data with a sampling rate of 1MHz, including signals such as human respiration (0.3Hz) and movement (1-3Hz).

[0039] S22. Perform CFAR (Constant False Alarm Rate) detection on the original ADC signal. Use an adaptive threshold to filter out environmental noise and clutter interference (such as static reflections from furniture, electromagnetic interference, etc.) to select effective signals related to human activity, ensuring that subsequent analysis focuses on real human targets. For example, use the CFAR algorithm to set an adaptive threshold (based on the average environmental noise of the previous 100ms) to filter static reflections from furniture (signals below the threshold) and retain effective signals of human activity to eliminate or reduce false triggers caused by environmental interference.

[0040] S23. Perform a distance-angle fast Fourier transform on the filtered signal to convert the time-domain signal into frequency-domain data, and analyze the spatial location information including the target's distance (distance from the radar) and angle (position relative to the radar) to generate spatial point cloud data (converting the human target into "distance-angle" coordinate points) to construct a spatial location framework of the human target in the guest room.

[0041] S24. Based on the generated spatial point cloud data, the DBSCAN point cloud clustering algorithm (density clustering algorithm) is used to cluster the discrete target point clouds (sets of human body reflection signals) in space into independent individuals, achieving multi-target differentiation, and then counting the real-time number of people in the guest rooms to obtain occupancy statistics. The DBSCAN point cloud clustering algorithm is a density-based spatial clustering algorithm, which can be implemented by setting a neighborhood radius and a minimum point count threshold to distinguish the point cloud distribution of different individuals. For example, by using the DBSCAN algorithm to cluster each frame of point cloud, setting the minimum number of clustered points to 5 (corresponding to human body contours), and setting the Euclidean distance threshold to 0.5m, two independent point cloud clusters are distinguished, and the number of people counted is 2. The Euclidean distance threshold is the critical distance value used in the DBSCAN density clustering algorithm to determine whether "two points belong to the same neighborhood," and its physical meaning is the upper limit of the straight-line distance between two points (or point clouds) in space. Specifically, when the straight-line distance between two point clouds is less than or equal to the Euclidean distance threshold (e.g., 0.5m), the algorithm determines that the two belong to the same human body (because the distance between different parts of the human body is usually less than 0.5m); when the straight-line distance between two point clouds is greater than the Euclidean distance threshold, the algorithm determines that the two belong to different human bodies (because there are obvious gaps between different human bodies).

[0042] S25: While S24 is being executed, Doppler phase analysis is performed based on the generated spatial point cloud data. Phase difference information is extracted using the Doppler effect (signal frequency change caused by human movement), the human movement direction vector is calculated, and the human movement trajectory and direction are analyzed. For example, based on the inter-frame displacement of the point cloud, the Doppler phase difference is calculated to obtain the movement direction vector (such as along the positive x-axis with a speed of 0.4 m / s), and the analyzed trajectory is "moving in a straight line from the activity area to the bathroom area".

[0043] S26: Integrate spatial location information, people statistics, and movement direction vectors to form real-time sensing data containing personnel presence status, real-time number of people, and movement direction information, providing input for device linkage triggering and scene control in step S3. The integrated data is structured information: {Presence status: Person present, Number of people: 2, Movement direction: (x: 0.8, y: 0.2)}.

[0044] As described above, this case utilizes CFAR constant false alarm rate detection to effectively filter interference signals such as furniture reflections and curtain swaying, improving the accuracy of human activity signal extraction and avoiding equipment malfunctions caused by noise misjudgments. For example, it prevents curtain swaying from being mistaken for human activity, thus improving detection reliability. The DBSCAN clustering algorithm achieves multi-target separation, enabling simultaneous identification of stationary individuals in the sleeping area and moving individuals in the activity area, avoiding misjudging stationary targets as unoccupied. Furthermore, by combining Doppler phase analysis and point cloud clustering data, it can simultaneously acquire information on the number, location, and direction of movement of individuals, supporting accurate identification when multiple people are in the room and providing data support for zoning control in multi-occupancy scenarios. For example, it can automatically adjust lighting and air conditioning modes when guests move from the sleeping area to the activity area, solving the scene switching delay problem caused by the inability to identify the direction of movement in existing technologies. Thus, the above multi-step signal processing (acquiring the raw ADC signal, CFAR filtering, FFT transformation, and point cloud clustering) solves the problems of high signal noise and low recognition accuracy in traditional sensing technologies.

[0045] like Figure 1 As shown, in one specific implementation method, step S2 of the guest room power supply method in this case further includes: S27. Based on the real-time sensing data, perform in-depth analysis of pedestrian flow direction and behavior prediction, specifically including: S271. Based on the spatial location information and movement direction vector in the real-time sensing data, assign a spatial location weight to the detected target at a certain location according to the preset weight value based on the spatial location information. Further analyze the movement speed of the detected target based on the movement direction vector and generate the target velocity vector by combining the movement direction vector. The target velocity vector comes from the movement direction vector and velocity parameters analyzed in S25. The spatial location weight is set according to the functional importance of key areas of the guest room (e.g., the weight of the doorway area is higher than that of other areas).

[0046] S272. Perform dynamic entropy analysis and calculate the movement trend value: The movement trend value is calculated as the sum of the products of the target velocity vector and the spatial position weight; as shown in the formula: Movement Trend = Σ(Target Velocity Vector × Spatial Position Weight). Here, the spatial position weight refers to a numerical coefficient set according to the functional importance of the guest room area, which can be implemented using a preset weight allocation table to reflect the degree of influence of personnel activities in different areas on equipment control. The target velocity vector is a two-dimensional vector parameter formed by combining movement speed and direction, which can be calculated using Doppler phase analysis to quantify the movement state of personnel. Dynamic entropy analysis refers to assessing the randomness of the movement trend through information entropy theory, which can be implemented by calculating the variance of the movement trend value to determine whether personnel activities have regularity.

[0047] S273. Construct a behavior prediction model. Calculate the rate of change of movement trend values ​​over 5 consecutive time frames (100ms interval between each frame) to obtain the trajectory entropy value. When the trajectory entropy value is greater than the preset threshold γ, it is determined to be a highly random movement, i.e., an unexpected chaotic movement (which may be an emergency), and triggers the emergency evacuation tag. When the trajectory entropy value is less than or equal to the preset threshold γ, it is determined to be a regular activity, that is, a regular activity that meets expectations, and the energy-saving scenario tag is triggered; (including the prediction of unmanned areas in the next 30 seconds and the pre-shutdown command of electrical equipment). The dynamic entropy analysis results and the tags output by the behavior prediction model are integrated into the real-time sensing data to provide a more refined decision-making basis for activating the device linkage list in step S3. Specifically, the emergency evacuation tag is an identifier that marks highly random movement and is used to trigger safety-related device linkages. The energy-saving scenario tag is an identifier that marks regular activities and is used to initiate energy-saving strategy control. In practical implementation, a spatial location weight of 1.2 can be assigned to the bathroom area (higher than the 1.0 of the activity area), and a target velocity vector (0.4 × 1.2 = 0.48) can be generated based on the movement direction vector (0.4 m / s, towards the bathroom area). The movement trend value = Σ (velocity vector of each target × position weight) = 0.48 (single person) + 0.36 (another person moves towards the sleeping area) = 0.84. The movement trend values ​​for 5 consecutive time frames (100ms each) are: [0.84, 0.86, 0.85, 0.87, 0.86]. The calculated rate of change is less than 0.02, and the trajectory entropy value = 0.3 (< threshold γ = 1.2). It is determined to be a regular activity, and an energy-saving scene label is generated.

[0048] As mentioned above, compared with existing technologies, current solutions can only determine the presence of personnel and cannot analyze movement trends and behavioral patterns. This solution, by introducing dynamic entropy analysis and trajectory entropy calculation, can effectively distinguish between routine activities and abnormal behaviors, and achieve differentiated equipment control by combining regional weights. For example, existing systems cannot identify emergency situations caused by rapid personnel movement, while this solution can promptly trigger safety linkage mechanisms through trajectory entropy monitoring.

[0049] Through the above technical solution, this application can accurately identify the movement behavior characteristics of people in guest rooms, solving the problem of misjudging static states by traditional technologies. Dynamic entropy analysis is used to classify behavior patterns, improving the accuracy of scene mode switching. Combined with a trajectory entropy threshold judgment mechanism, energy management efficiency is optimized while ensuring user experience, avoiding equipment malfunctions caused by single-state detection.

[0050] As one specific implementation, step S3 includes: Step S3 includes: S31. Receive the real-time sensing data output in step S2 in real time, and extract human activity feature parameters in each sensing area. The human activity feature parameters include human movement speed, dwell time, target quantity change rate and activity area coverage. S32. Compare and analyze the extracted human activity feature parameters with the pre-set trigger conditions for the sensing area: When the human body's movement speed is greater than the first speed preset threshold and the duration is greater than or equal to the first time preset threshold, it is determined to be "dynamic activity triggered"; When the human body's movement speed is less than or equal to the first speed preset threshold and the duration is greater than or equal to the second time preset threshold, it is determined to be "static dwell trigger". When the number of targets in the sensing area increases from zero, it is determined as "new entry triggered"; When the number of targets in all sensing areas decreases from present to absent, it is determined as "departure trigger". Specifically, when the human movement speed is >0.3m / s and the duration is ≥2s, it is determined as "dynamic activity trigger"; when the human movement speed is ≤0.3m / s but the dwell time is ≥5s, it is determined as "static dwelling trigger"; when the number of targets in the sensing area increases from 0 to ≥1, it is determined as "new entry trigger"; when the number of targets in all sensing areas changes from ≥1 to 0, it is determined as "departure trigger".

[0051] S33. If any of the above triggering conditions are met, the central control module will call the pre-configured device linkage list of the sensing area through the network module. The device linkage list includes the trigger priority order (lighting adjustment > air conditioning control > curtain linkage) and parameter thresholds (such as lighting brightness adjustment range, air conditioning temperature setting range). S34. For sensing areas that do not meet the trigger conditions, maintain the current device operating state and continuously monitor changes in their characteristic parameters. If no human activity is detected for 30 consecutive seconds, it is marked as a low-activity area. For example, extract the characteristic parameters of the activity area: movement speed 0.6m / s, dwell time 10s, target quantity change rate +1 (from 1 person to 2 people), and activity coverage rate 60%. Because the movement speed > 0.3m / s (first speed threshold) and lasts for 10s ≥ 2s (first time threshold), it is determined to be "dynamic activity triggered". Call the device linkage list of the activity area, with the priority order as "lighting > TV > curtains", and the parameter thresholds as follows: lighting brightness 50%-100%, TV volume 0-50dB. If no activity is detected in the bathroom area for 30 consecutive seconds, it is marked as a low-activity area.

[0052] As mentioned above, existing technologies can only achieve basic manned / unmanned status detection, and suffer from numerous problems such as susceptibility to misjudgment and a lack of regional collaborative judgment capabilities. This solution, however, achieves precise triggering and on-demand response for device linkage through multi-dimensional triggering conditions (dynamic / static / new entry / exit). On one hand, differentiated triggering based on parameters such as human movement speed and dwell time effectively avoids misjudgments that may occur with single-condition judgments, accurately matching device linkage needs in different scenarios. Simultaneously, a low-activity area = marking mechanism ensures timely shutdown of devices in unmanned areas, significantly reducing energy consumption and achieving more efficient energy management compared to traditional "fixed-duration shutdown" solutions. On the other hand, binding triggering conditions to the priority of the device linkage list ensures priority response from critical devices, thereby improving the smoothness of the user experience and avoiding inconvenience caused by chaotic device response sequences. In addition, this solution constructs four refined triggering modes. By setting a dual time threshold mechanism, it can accurately distinguish between brief periods of stillness and true unmanned state, effectively avoiding misjudgment caused by brief periods of stillness. At the same time, through a global target quantity monitoring mechanism, it can achieve accurate departure trigger judgment, solving the problem of overall power outage caused by misjudgment in a single area.

[0053] Through the above technical solutions, this application solves the technical defects of traditional technologies, such as high misjudgment rate of personnel status and inability to identify static states, and realizes multi-dimensional perception of personnel activities in guest rooms. By setting dual judgment conditions of dynamic activity and static stay, the problem of accidental device shutdown during sleep is avoided. Through a globally collaborative target quantity monitoring mechanism, the accuracy of departure triggering in multi-person scenarios is ensured, effectively preventing unnecessary power outages caused by temporary departures of personnel. The low-activity area marking function further optimizes the energy management strategy, reducing energy consumption in inactive areas while ensuring basic functions.

[0054] As a specific implementation method, this application further proposes that the "power supply control command" in step S4 include the following generation mode: Continuous power supply mode: In response to "dynamic activity trigger" and "new entry trigger" states, a continuous power supply command is generated to keep the guest room power supply system on. Energy-saving power supply mode: In response to the "static dwell trigger" state, an energy-saving power supply command is generated, which automatically reduces the power supply of smart electrical devices outside the sensing area; Delayed power-off mode: In response to the "away trigger" state, a delayed power-off command is generated, and after a preset delay, non-essential power supply is cut off, while power supply to critical equipment is preserved.

[0055] As mentioned above, compared with existing technologies, current solutions have significant limitations. They can only power on or off the entire area based on a single presence signal, and cannot dynamically adjust the power supply to different zones based on the actual activity status of personnel. For example, traditional radar systems often misjudge a static human body as being unoccupied, leading to abnormal equipment shutdowns and severely impacting user experience. This solution innovatively employs a multi-mode power supply command generation mechanism, enabling differentiated control of equipment in different areas based on various trigger conditions. This effectively avoids the problem of accidental equipment shutdowns caused by relying on a single human status identification, greatly improving operational accuracy. Furthermore, the optimized zoned power supply strategy in multi-person scenarios significantly improves energy efficiency.

[0056] Specifically, this solution demonstrates unique advantages in different scenarios. In static stay scenarios, by reducing the power of equipment in inactive areas, energy consumption is effectively reduced while fully ensuring user experience; the delayed power-off mechanism avoids repeated start-ups and shutdowns of equipment when people temporarily leave, enhancing system stability and extending equipment lifespan. Furthermore, this solution sets up three power control modes to precisely adapt to different scenarios, balancing "power supply reliability" and "energy efficiency." The continuous power supply mode ensures that guests will not encounter power outages upon arrival, solving the "card insertion delay" problem inherent in traditional card-based power supply; the energy-saving power supply mode reduces power in unnecessary areas during single-person sleep scenarios, reducing energy consumption without affecting user rest; the delayed power-off mode prevents equipment shutdown due to guests temporarily leaving the room, while ensuring that unnecessary power is cut off when the room is unoccupied for extended periods (while retaining critical equipment such as refrigerators and emergency lighting), further reducing ineffective energy consumption.

[0057] As a specific implementation method, this application further proposes that the "scene mode switching instruction" in step S4 is a specific execution instruction for the device linkage list activated in step S3, including preset modes based on the activity status of personnel in different areas: Welcome Mode: The device linkage list corresponding to the "new entry trigger" in the entrance area is used to control the lighting, curtains, and air conditioning equipment to switch to the state that the person has just entered; for example, when the "new entry trigger" in the entrance area, the corridor lights gradually brighten from 10% to 80% brightness (3s), the curtains open to 70%, the air conditioning is set to 25℃, and the smart speaker plays a welcome message.

[0058] Sleep Mode: A list of devices linked to the "static stay trigger" in the sleep area to control lighting, curtains, and air conditioning to switch to a state suitable for rest; for example, in sleep mode: when the sleep area is "static stay triggered" (20 minutes of stillness) and there is no one in the activity area, the main light is turned off, the bedside lamp is adjusted to 10% brightness, the air conditioner switches to silent mode, and the curtains automatically close to a completely dark state.

[0059] Activity Mode: The device linkage list corresponding to the activity area's "Dynamic Activity Trigger" controls the switching of lighting and entertainment equipment to an operating state suitable for multi-person interaction or activities; for example, when the main lighting is adjusted to 80% brightness (white light mode, color temperature 5000K), the TV automatically turns on and switches to standby mode, the background music system starts (playing light music, volume 30dB), and if ≥4 people are detected, the fresh air system is automatically turned on (air exchange frequency increased to 6 times per hour).

[0060] Bathroom Mode: This mode controls the bathroom lighting and ventilation equipment to switch to a state suitable for washing and bathing scenarios when the device is triggered by "new entry" or "static stay". For example, the vanity light is 100% on (color rendering index ≥ 90), the exhaust fan is started (air volume 150m³ / h), and if the stay time is detected to be ≥ 5 minutes (determined to be a bathing scenario), the bedroom lighting linked to the bathroom door magnet is automatically turned off (to avoid strong light interference).

[0061] Energy-saving mode: For the device linkage list marked "low activity area", control the power of non-essential electrical equipment in the area to reduce or turn off to match the low activity demand; for example, the power of lighting equipment is reduced to 30% (or non-essential lighting is turned off), the air conditioner is switched to energy-saving mode (adjusted to 28°C in summer and 20°C in winter), entertainment equipment (such as TV and stereo) automatically enters standby mode, and smart sockets cut off the power supply to unloaded equipment such as chargers.

[0062] Leave Mode: This mode controls the devices linked to the "Leave Trigger" list in the entrance area, switching them to the appropriate state for when someone leaves. For example, all lights turn off sequentially (corridor lights turn off last, with a 5-second delay), the air conditioner switches to ventilation mode (running for 10 minutes per hour), the curtains remain in their current state (or are closed to 50% by default), the smart door lock automatically locks, and the system sends a "Pending Inspection" notification to the housekeeping department.

[0063] As mentioned above, compared with existing technologies, current solutions have many shortcomings. They can only control device switching based on a single presence signal, failing to implement differentiated scene control based on regional characteristics and activity types. Furthermore, traditional solutions lack precision in device control. For example, in bathroom areas, traditional solutions may misjudge and cause frequent start-stop cycles for ventilation equipment, resulting in noise interference and shortening equipment lifespan. This solution, through regionalized mode division, allows ventilation equipment to activate only when personnel are detected entering. In addition, existing technologies lack a mechanism for linking entertainment equipment in multi-person activity scenarios, failing to flexibly adjust equipment combinations according to actual activity conditions. This solution, however, can automatically adjust equipment combinations based on dynamic activity intensity to meet the needs of different scenarios.

[0064] This application, through the aforementioned technical solution, solves the problems of traditional guest room control scenarios being monotonous and lacking regional specificity in equipment linkage. By dividing the room into regional modes and establishing preset modes that match the activity characteristics of key areas, personalized experience upgrades are achieved. For example, the welcome mode creates a comfortable room entry experience, and the sleep mode enhances the nighttime rest experience. Equipment control is precise and intelligent, with bathroom ventilation and other equipment being accurately triggered and dynamically adjusted. Energy-saving effects are significant, with energy-saving modes and departure modes reducing energy consumption. The system is stable and reliable, avoiding malfunctions due to misoperation and fully ensuring functional operation, showing clear advantages over traditional solutions.

[0065] In one specific implementation, the radar sensor is a 60GHz human presence detection radar sensor. Specifically, the radar sensor uses the Infineon BGT60UTR11AIP radar sensor, which measures 4mm × 4mm, making it extremely compact and the smallest 60GHz radar sensor with an integrated antenna on the market. Thus, this sensor operates in the 60GHz frequency band, accurately detecting human presence and providing reliable human activity information to the system. Its small size significantly saves installation space, reduces requirements for equipment layout, and allows the sensor to be flexibly applied in various space-constrained scenarios. It also helps improve the overall integration and aesthetics of the device, providing strong support for the stable operation and efficient application of the system.

[0066] like Figure 2 As shown, in one specific implementation, the radar sensor includes: a Fresnel lens beamforming unit 11, a MIMO antenna array 12, and an algorithm processing unit 13; Among them, the Fresnel lens beamforming unit 11 enhances the directional reception capability of spatial presence sensing signals, thereby enhancing the micro-motion signals generated by human life activities, so as to realize the recognition of the static / active state of the human body. The MIMO antenna array 12 is used to collect multi-dimensional human movement trajectory signals and calculate the human body azimuth angle by the phase difference of the multi-channel received signals to obtain the human body's azimuth information relative to the radar. The algorithm processing unit 13 loads a multi-target point cloud separation algorithm and a neural network model. The multi-target point cloud separation algorithm performs density clustering on the spatial point cloud data collected by the MIMO antenna array 12, distinguishes different human targets by setting a distance threshold, and completes the people counting. The neural network model receives continuous azimuth angle data, constructs a people flow line model, and outputs a heat map of people activity and trajectory prediction results, providing a basis for people movement for power supply control. Specifically, the Fresnel lens beamforming unit: enhances the reception intensity of human chest cavity micro-movement (0.02m / s) by focusing radar signals through the lens, and distinguishes stationary human bodies from still objects. The MIMO antenna array (8 channels) calculates the human body azimuth angle (accuracy ±3°) through phase difference and outputs the movement trajectory coordinates in real time (e.g., (x:3.2m, y:1.5m)). The algorithm processing unit runs the DBSCAN algorithm (distance threshold 0.6m) to distinguish 3 target point clouds, and the LSTM model predicts the trajectory based on 200 frames of azimuth angle data (e.g., "moving towards the door after 10s").

[0067] The Fresnel lens beamforming unit refers to a component that focuses and beamforms radar waves using a Fresnel lens structure. Specifically, it can be implemented using a lens with a stepped concentric ring structure. By altering the propagation path of electromagnetic waves, it enhances the signal reception sensitivity in a specific direction, thereby improving the ability to capture subtle human movement signals in a stationary state. The MIMO antenna array is a signal acquisition module employing a multiple-input multiple-output antenna structure. Specifically, it can be implemented using an array layout consisting of a four-channel transmitting antenna and a six-channel receiving antenna. It calculates the target azimuth angle using the phase difference of multi-channel signals to obtain precise spatial location information of the human body. The algorithm processing unit is a computational module integrating data processing algorithms. Specifically, it can be implemented using an embedded processor loaded with density clustering algorithms and recurrent neural network models. It distinguishes multiple targets and predicts movement trajectories through point cloud clustering, supporting population counting and dynamic behavior analysis.

[0068] Specifically, the Fresnel lens beamforming unit enhances the detection sensitivity of micro-motion signals such as chest cavity breathing fluctuations by focusing 60GHz high-frequency electromagnetic waves, enabling the effective identification of vital signs in a static state. The MIMO antenna array acquires phase difference data of reflected signals at different angles through multi-channel signal transmission and reception, and combines this with a direction-of-arrival (DOA) estimation algorithm to analyze the human azimuth angle and form spatial coordinate information. The algorithm processing unit performs density-based clustering analysis on the raw point cloud data, dividing spatially discrete points into independent individuals by setting distance thresholds to accurately count the number of people in the guest rooms; simultaneously, it uses a neural network model to perform time-series analysis on continuous azimuth angle data, constructing a personnel movement trend model and outputting heat maps and trajectory prediction results, providing dynamic basis for equipment linkage.

[0069] As described above, the radar sensor used in this application significantly enhances the overall perception capability of "hardware + algorithm" through the collaborative work of its various units, bringing about several outstanding and beneficial effects. At the hardware level, the Fresnel lens, by enhancing the reception of micro-motion signals, effectively solves the problem of inaccurate identification of stationary human bodies and the tendency to "misjudge stillness" in traditional radar, accurately distinguishing between the stationary and active states of a human body. The MIMO antenna array overcomes the limitations of traditional single-antenna radar azimuth detection, improving azimuth accuracy through multi-channel phase difference calculation, reliably tracking trajectories, and supporting the differentiation of individuals when multiple people are moving together, providing strong data support for zoning control in residential settings. At the algorithm level, the algorithm processing unit combines point cloud separation with a neural network model, overcoming the shortcomings of existing technologies in counting people and predicting trajectories. It can not only accurately count people but also predict trajectories. Compared with existing millimeter-wave radar technologies that can only achieve basic presence detection, the 60GHz radar sensor of this application, through the above innovative design, provides refined data support for the linkage of scenario-based devices. Based on this, this application effectively avoids the situation where equipment is accidentally shut down due to misjudgment, realizes accurate number of people and zoned equipment control in multi-person scenarios, and can also generate heat maps based on trajectory prediction, providing a basis for dynamic adjustment of equipment linkage strategies in different areas, greatly improving the accuracy and response efficiency of guest room scene mode switching.

[0070] Through the above technical solutions, this application effectively solves the problem of insufficient accuracy of traditional radar sensors in recognizing stationary human bodies, avoiding accidental shutdown of equipment due to misjudgment; it achieves accurate number of people statistics through multi-target point cloud separation, supporting zoned equipment control in multi-person scenarios; and the heat map generation function based on trajectory prediction provides data basis for dynamic adjustment of equipment linkage strategies in different areas, improving the accuracy and response efficiency of guest room scene mode switching.

[0071] As a preferred implementation, step S27, the in-depth analysis of pedestrian flow direction and behavior prediction, further includes the following steps: Real-time sensing data is uploaded to the SAAS system via a network module. The SAAS system then calls a preset big data analysis model to jointly analyze the historical sensing data and real-time data, optimizing the trajectory entropy calculation parameters and preset thresholds in the behavior prediction model. The big data analysis model dynamically adjusts the judgment logic under different scenarios based on the activity feature samples of multiple guest rooms, so that the accuracy of distinguishing between highly random movement and regular activities is adapted to the actual use scenario of the guest rooms.

[0072] The SaaS system refers to a software service system based on a cloud computing architecture, specifically implemented using a distributed data storage and computing framework. It integrates historical and real-time sensor data from multiple guest rooms, providing cross-regional data analysis capabilities. The big data analysis model refers to a predictive model trained using machine learning algorithms. Specifically, after initial analysis of the real-time sensor data, the network module uploads the data to the SaaS system. The SaaS system calls upon stored historical data, combining it with current personnel movement trends, trajectory entropy values, and trigger tags within the guest rooms, to calculate the optimal parameter combinations for different scenarios using the big data analysis model. For example, in business hotel scenarios, personnel movement trajectories typically exhibit high regularity; in this case, the model automatically lowers the threshold for judging trajectory entropy values ​​to avoid misjudging rapid walking as highly random movement. In family-themed guest rooms, the model dynamically raises the judgment threshold based on the high-frequency random characteristics of children's activities in historical data, ensuring accurate differentiation between regular activities and abnormal behavior. Through continuous iterative optimization, the behavior prediction model can adapt to different guest room types and improve the reliability of detecting highly random movement. In practice, the network module uploads real-time sensing data (once every 5 minutes) to the SAAS system via WiFi; the system calls the big data model (based on 100,000 historical data from 100 guest rooms) to optimize the trajectory entropy threshold: the γ value for family suites is adjusted from 1.2 to 1.5 (to adapt to the high randomness of children running), and the γ value for business rooms is reduced to 1.0 (to reduce false positives), thereby improving the accuracy of abnormal behavior identification.

[0073] As mentioned above, compared with existing technologies, traditional solutions rely solely on real-time data from a single guest room to determine behavior, failing to utilize historical data from across guest rooms to optimize the model. This results in rigid judgment logic, making it difficult to adapt to the differentiated needs of different scenarios and leading to a high false positive rate in complex situations. This application, however, leverages a SaaS system to integrate multi-source data, enabling the behavior prediction model to possess self-learning capabilities and dynamically adjust judgment parameters according to the actual scenario, effectively solving the aforementioned problems. This solution achieves accurate identification and adaptive optimization of personnel activity patterns, overcoming the poor scenario adaptability of existing technologies. Specifically, cross-guest room data joint analysis improves the accuracy of distinguishing between highly random movements and routine activities, avoiding false triggers caused by frequent running by children in parent-child scenarios; dynamically adjusting trajectory entropy parameters makes energy-saving scenario label generation more realistic, optimizing the energy management efficiency of guest room equipment. Big data optimization in SaaS systems has driven the upgrade from "single-room experience" to "multi-room intelligence." It not only dynamically adjusts trajectory entropy thresholds based on different room types (such as family suites) to improve the accuracy of abnormal behavior identification, but also utilizes historical data to optimize behavior prediction model parameters, reducing misjudgments caused by differences in room layouts and significantly improving system adaptability. Furthermore, energy consumption heatmaps generated from cross-room data analysis help hotels identify high-energy-consuming areas, guide operational optimization, and effectively reduce overall energy consumption.

[0074] The foregoing has provided a detailed description of a method for obtaining power to guest rooms based on radar identification of the number and movement of people in a space, as disclosed in the embodiments of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for controlling power supply in guest rooms based on radar-based identification of the number and movement of people in the space, characterized in that, include: Step S1: Obtain the layout plan of the target guest room, identify key areas of the guest room, and configure one or more smart switches according to the key areas of the guest room, while dividing the sensing areas; wherein, the smart switch includes at least: a radar sensor, a central control module, and a network module, and each smart switch in the sensing area is configured with a corresponding device linkage list; Step S2: The smart switch collects spatial detection signals in the guest room in real time through radar sensors and generates real-time sensing data. It also processes the spatial detection signals and real-time sensing data to obtain information on the presence status of people in the guest room, the real-time number of people, and the direction of movement. The spatial detection signals include human vital signs signals, movement trajectory signals, and multi-target distribution signals. Step S3: When the real-time sensing data of a certain sensing area meets the preset trigger conditions, activate the device linkage list corresponding to that area. Step S4: Generate a power supply control command based on the personnel presence status, real-time number of people, and movement direction information, and generate a scene mode switching command based on the activated device linkage list; Step S5: Control the on / off state of the guest room power supply system based on the power supply control command, and switch the smart electrical equipment in the guest room to the corresponding operating mode according to the scene mode switching command.

2. The method for obtaining power to a guest room according to claim 1, characterized in that, The key areas of the guest room mentioned in step S1 include: the entrance area, the sleeping area, the activity area, and the bathroom area; the sensing area is the detection range formed according to the scanning angle of each smart switch.

3. The method for obtaining power to a guest room according to claim 1, characterized in that, Step S2 includes: S21. The radar sensor emits electromagnetic waves and receives reflected signals to obtain the raw ADC signal as the basic data for space detection, covering the initial signals related to human vital signs, movement trajectories and multi-target distribution in the guest room. S22. Perform CFAR constant false alarm rate detection on the original ADC signal and filter out environmental noise and clutter interference through adaptive threshold to select effective signals related to human activities. S23. Perform a fast Fourier transform on the filtered signal to convert the time domain signal into frequency domain data, analyze the spatial location information including the distance and angle of the target and generate spatial point cloud data to construct a spatial location framework of the human target in the guest room. S24. Based on the generated spatial point cloud data, the DBSCAN point cloud clustering algorithm is used to cluster the discrete target point clouds in space into independent individuals to achieve multi-target differentiation, thereby counting the real-time number of people in the guest rooms to obtain the number of people statistics information. S25. While S24 is being executed, Doppler phase analysis is performed based on the generated spatial point cloud data. Phase difference information is extracted using the Doppler effect, the human body's movement direction vector is calculated, and the human body's movement trajectory and direction are analyzed. S26. Integrate spatial location information, people statistics, and movement direction vectors to form real-time sensing data that includes people's presence status, real-time number of people, and movement direction information.

4. The method for obtaining power to a guest room according to claim 3, characterized in that, Step S2 also includes: S27. Based on the real-time sensing data, perform in-depth analysis of pedestrian flow direction and behavior prediction, specifically including: S271. Based on the spatial position information and movement direction vector in the real-time sensing data, assign a spatial position weight to the detected target at a certain position according to the spatial position information and a preset weight value, and further analyze the movement speed of the detected target according to the movement direction vector and generate the target velocity vector by combining the movement direction vector. S272. Perform dynamic entropy analysis and calculate the movement trend value: The movement trend value is calculated as the sum of the products of the target velocity vector and the spatial position weight. S273. Construct a behavior prediction model and obtain the trajectory entropy value by calculating the rate of change of the movement trend value over n consecutive time frames. When the trajectory entropy value is greater than the preset threshold, it is judged as a highly random movement and an emergency evacuation tag is triggered. When the trajectory entropy value is less than or equal to a preset threshold, it is determined to be a regular activity and an energy-saving scenario label is generated. The trajectory entropy analysis results and output labels are integrated into the real-time sensing data.

5. The method for obtaining power to a guest room according to claim 1 or 4, characterized in that, Step S3 includes: S31. Receive the real-time sensing data output in step S2 in real time, and extract human activity feature parameters in each sensing area. The human activity feature parameters include human movement speed, dwell time, target quantity change rate and activity area coverage. S32. Compare and analyze the extracted human activity feature parameters with the pre-set trigger conditions for the sensing area: When the human body's movement speed exceeds the first speed preset threshold and the duration is greater than or equal to the first time preset threshold, it is determined to be "dynamic activity triggered"; When the human body's movement speed is less than or equal to the first speed preset threshold and the duration is greater than or equal to the second time preset threshold, it is determined as "static dwell trigger"; When the number of targets in the sensing area increases from zero, it is determined as "new entry triggered"; When the number of targets in all sensing areas decreases from present to absent, it is determined as "departure trigger"; S33. If any of the above triggering conditions are met, the central control module calls the pre-configured device linkage list of the sensing area through the network module. The device linkage list includes the trigger priority order of each smart electrical device and the relevant parameter threshold settings. S34. For sensing areas that do not meet the trigger conditions, maintain the current operating state of the device and continuously monitor changes in its characteristic parameters. When no human activity characteristics are detected for a continuous period of time, mark it as a low-activity area.

6. The method for obtaining power to a guest room according to claim 5, characterized in that, The "power control command" in step S4 includes the following generation modes: Continuous power supply mode: In response to "dynamic activity trigger" and "new entry trigger" states, a continuous power supply command is generated to keep the guest room power supply system on. Energy-saving power supply mode: In response to the "static dwell trigger" state, an energy-saving power supply command is generated, which automatically reduces the power supply of smart electrical devices outside the sensing area; Delayed power-off mode: In response to the "away trigger" state, a delayed power-off command is generated, and after a preset delay, non-essential power supply is cut off, while power supply to critical equipment is preserved.

7. The method for obtaining power to a guest room according to claim 5, characterized in that, In step S4, the "scene mode switching command" is a specific execution command for the device linkage list activated in step S3, including preset modes based on the activity status of people in different areas: Welcome Mode: This mode controls the lighting, curtains, and air conditioning equipment to switch to the state appropriate for when a person enters the area, based on the "new entry trigger" device list in the entrance area. Sleep Mode: A list of devices linked to the "static stay trigger" in the sleep area to control lighting, curtains, and air conditioning to switch to a state suitable for rest; Activity Mode: The device linkage list corresponding to the "Dynamic Activity Trigger" in the activity area, to control the lighting and entertainment equipment to switch to the operating state that is suitable for multi-person interaction or activities; Bathroom Mode: This mode controls the bathroom lighting and ventilation equipment to switch to a state suitable for washing and bathing scenarios, based on the device linkage list triggered by "new entry" or "static stay" in the bathroom area. Energy Saving Mode: For the device linkage list marked "Low Activity Area", control the power of non-essential electrical equipment in the area to reduce or turn off to match the low activity demand; Leave Mode: The list of devices linked to the "Leave Trigger" in the corresponding doorway area, controlling the devices to switch to the state where the personnel are leaving.

8. The method for obtaining power to a guest room according to claim 1 or 3, characterized in that, The radar sensor is a 60GHz human presence sensing radar sensor.

9. The method for obtaining power to a guest room according to claim 8, characterized in that, The radar sensor includes: a Fresnel lens beamforming unit (11), a MIMO antenna array (12), and an algorithm processing unit (13). Among them, the Fresnel lens beamforming unit (11) enhances the directional reception capability of spatial presence sensing signals, thereby enhancing the micro-motion signals generated by human life activities, so as to realize the recognition of the static / active state of the human body. The MIMO antenna array (12) is used to collect multi-dimensional human movement trajectory signals and calculate the human azimuth angle by the phase difference of the multi-channel received signals to obtain the human azimuth information relative to the radar. The algorithm processing unit (13) loads a multi-target point cloud separation algorithm and a neural network model; the multi-target point cloud separation algorithm performs density clustering on the spatial point cloud data collected by the MIMO antenna array (12), distinguishes different human targets by setting a distance threshold, and completes the number of people count; the neural network model receives continuous azimuth angle data, constructs a human flow line model, outputs a human activity heat map and trajectory prediction results, and provides a basis for human movement for power supply control.

10. The method for obtaining power to a guest room according to claim 4, characterized in that, Step S27, the in-depth analysis of pedestrian flow and behavior prediction, also includes the following steps: Real-time sensing data is uploaded to the SAAS system via a network module. The SAAS system then calls a preset big data analysis model to jointly analyze the historical sensing data and real-time data, optimizing the trajectory entropy calculation parameters and preset thresholds in the behavior prediction model. The big data analysis model dynamically adjusts the judgment logic under different scenarios based on the activity feature samples of multiple guest rooms, so that the accuracy of distinguishing between highly random movement and regular activities is adapted to the actual use scenario of the guest rooms.

Images