Bracelet tumble detection system and detection method based on magnetic attraction split type framework

Abstract

The invention discloses a bracelet tumble detection system and detection method based on a magnetic attraction split type framework, and belongs to the technical field of intelligent wearing. The system comprises an upper watch body and a lower watch body which can be separated, and the lower watch body continuously monitors the connection state of an acceleration sensor and a magnetic attraction contact. When a user falls down, a transient signal interaction change of a contact can be caused by a violent impact force, and a specific electric signal characteristic is generated. According to the invention, through fusion analysis of the acceleration sensor threshold and the contact transient signal, high-precision and low-false-alarm identification of the fall event is realized. Meanwhile, the split type design is matched with a double-battery management strategy, so that the problem of monitoring interruption during the charging period is solved. The system also has a multi-level alarm link guarantee, and ensures reliable transmission of help-seeking information. According to the invention, the safety monitoring capability of wearable equipment of high-risk crowds such as old people is effectively improved.

Description

Technical Field

[0001] This invention relates to the field of smart wearable device technology, specifically to a wristband fall detection system and method based on a magnetic split architecture. Background Technology

[0002] Existing fall detection technologies for senior wristbands suffer from high false alarm rates, lack of monitoring during charging, and limited algorithm computing power. Traditional solutions rely on a single accelerometer, making it difficult to distinguish between falls and vigorous daily movements.

[0003] For example, patent number CN202310397669.8, entitled "An Invention Patent for a Smart Monitoring System for Fall Prevention," describes a fall prevention system based on video and body data, which provides early warnings by establishing a safe behavior model. However, such systems rely on external fixed monitoring equipment and lack the portability and real-time performance of wearable devices, making it difficult to solve the problem of false alarms from wearable devices, which is the focus of this application. The system includes a video monitoring module and a body monitoring module; it also includes a processor, an alarm, a storage device, and a learning chip. The processor processes historical monitoring videos and historical body data and stores them in the storage device. The learning chip retrieves historical video monitoring data from the storage device, establishes a safe route model, and stores the safe route in the storage device. The learning chip retrieves historical video monitoring data and body monitoring data of the target person from the storage device, establishes a safe behavior model, and stores the safe behavior model in the storage device. The processor compares the real-time monitoring data of the target person with the safe route model and the safe behavior model. When the target person deviates from the safe route model or the safe behavior model, the alarm is triggered.

[0004] Existing fall detection wristbands for the elderly face three major challenges: High false alarm rate: Relying on a single accelerometer, it cannot effectively distinguish between falling down and actions such as clapping, slapping a table, or waving, because the peak acceleration values ​​of these actions are often similar.

[0005] Charging gap: The wristband needs to be removed for charging, leaving the elderly without supervision at night (a high-risk period for falls).

[0006] Limited computing power of the algorithm: The MCU of the wristband control unit has limited computing power, making it difficult to run complex neural network models for data fusion and false alarm filtering.

[0007] While existing technologies incorporate barometers to detect changes in altitude, their effectiveness is limited when the fall height is small (such as slipping on flat ground). Therefore, there is an urgent need for a solution that introduces a new physical dimension, consumes low power, and is available in all weather conditions. Summary of the Invention

[0008] To address the problems existing in the prior art, the present invention aims to provide a wristband fall detection system and method based on a magnetic split architecture and multi-dimensional signal fusion, achieving high-precision fall determination and effectively filtering false alarms caused by vigorous movements such as waving and running.

[0009] To address the problems in the existing technology, the present invention adopts the following technical solution: A fall detection system for a smart bracelet based on a magnetically attached split architecture includes a smart bracelet, a mobile terminal, and a cloud alarm platform. The smart bracelet adopts a split upper and lower body architecture, comprising a separable upper body and a lower body. The physical connection surfaces of the upper and lower bodies are respectively equipped with a magnetically conductive mechanism and a mechanical limiting mechanism at both ends, achieving magnetic attraction and separation limiting of the upper and lower bodies. The upper body houses a first control unit (MCU), a display module, a touch module, and a main battery. The lower body houses a second control unit (MCU), an accelerometer, a backup battery, a wireless communication module, and an audible and visual alarm unit. The second control unit MCU of the meter body is equipped with a connection status monitoring circuit and a voice prompt module, which is used to monitor the physical characteristic signals of the conductive connection between the upper and lower meter bodies in real time. The lower meter body collects acceleration sensor and connection status signal data, and performs signal fusion and judgment through the second control unit MCU. The second control unit MCU simultaneously captures two signals: high G-value impact and circuit interruption, which constitute a dual signal of fall. When a fall is determined, the system activates a local alarm and sends a command to the mobile terminal via Bluetooth, and finally uploads it to the cloud platform. The upper meter body can be detached and completely separated from the lower meter body, while the lower meter body can continue to perform monitoring.

[0010] Furthermore, the magnetic attraction conductive mechanism includes multiple communication contacts and a permanent magnet, which are used to provide electrical and physical adsorption connections between the upper and lower watch bodies; the mechanical limiting mechanism is located on one side of the watch body, forming a hinged or snap-fit ​​connection, which is used to limit the maximum displacement distance of the upper watch body from the lower watch body.

[0011] Furthermore, the connection status monitoring circuit is configured to monitor at least one of the following signals to generate a physical characteristic signal of the conductive connection between the upper and lower body: (a) The electrical on or off state of a communication contact, regardless of whether the contact carries serial communication, parallel communication, I2C, SPI protocol or a simple level signal; (b) Changes in the magnetic field strength output by a magnetic field sensor located inside the body of the instrument; (c) Changes in the mutual inductance or capacitive coupling signal between the coils of the meter body; wherein the signal interaction change is caused by the relative displacement or transient separation of the upper and lower meter bodies due to external force impact.

[0012] Furthermore, the mechanical limiting mechanism is configured as a semi-separated protection structure. When the wristband is subjected to an impact force greater than the magnetic attraction threshold, the magnetic attraction conductive mechanism is disconnected, and the mechanical limiting mechanism forces the upper body to remain in a non-electrically loose state near the lower body.

[0013] Furthermore, the physical characteristic signal used for real-time monitoring of the conductive connection between the upper and lower body has three states: Steady-state connection: Signals interact normally, magnetic field strength is stable, or signal quantity is at the reference value; Physical separation: signal interruption, magnetic field strength below the separation threshold, or continuous abrupt change in signal quantity; Transient impact: The signal undergoes a 'on→off→on' transition within a preset time window, or the magnetic field strength / signal quantity undergoes a sudden change and then recovers.

[0014] Furthermore, a method for detecting a wristband falling includes the following steps: S1. Dual data stream monitoring: The lower body collects data from the accelerometer sensor, and at the same time collects and scans the signal interaction changes of the electrical connection status with the upper body; S2. Feature Fusion Judgment: When the collected combined acceleration sensor value exceeds the first preset threshold, the event judgment window is entered, and the fall judgment is made by combining the physical feature signal of the electrical connection between the upper and lower body. S3. Multimodal alarm: After confirming a fall, the system activates a local audible and visual alarm and sends a distress message to the mobile terminal via a wireless communication module, and finally reports it to the cloud alarm platform.

[0015] Furthermore, the feature fusion determination step includes: (1) Suspected fall or minor fall: Acceleration > 3G, and at the same time, the electrical connection status signal of the upper and lower body of the watch body shows a single "on → off → on" jump within a preset time, such as a pulse width of 10ms-500ms; at this time, a voice prompt asks whether to alarm; at this time, the alarm can be canceled by using the function button set on the wristband. Judgment: If no choice is made, the system will default to triggering an alarm signal after 3 seconds if the person has fallen. (2) Heavy fall: Acceleration > 5G, judgment: directly issue an alarm signal; (3) Vigorous exercise / clapping: acceleration > 3G and is periodic, while the electrical connection status signal of the upper and lower body shows a cyclic "on → off → on" jump or high frequency glitch; Judgment: Ignore.

[0016] Furthermore, the detection method also includes a motion interference filtering algorithm, which distinguishes between vigorous movement and a real fall by analyzing the connection state signal waveform, specifically: S1. When the accelerometer detects a continuous high G-value signal, the system analyzes the waveform of the connection status signal; S2. If the connection status signal exhibits periodic, regular microsecond-level high-frequency oscillations, it is determined to be mechanical resonance caused by strenuous exercise (such as running or clapping), and the system disables the fall alarm; S3. Only when the connection status signal exhibits a non-periodic, step-like millisecond-level disconnection characteristic is it considered a valid input for fall detection.

[0017] Furthermore, the alarm process includes a multi-level link protection mechanism, including Bluetooth connection, system-level background service monitoring, and local audible and visual alarms; specifically: Bluetooth connection: The lower part of the table connects to the mobile terminal APP via Bluetooth BLE, and the alarm is uploaded through the APP's data channel; Background service monitoring: If the primary link fails and the APP is detected and killed in the background, the table below switches to Bluetooth broadcast mode. After the system-level background service of the mobile terminal (such as the accessibility service of Android or the background process of iOS) hears the specific broadcast, it automatically wakes up the APP process and performs alarm uploading. Local audio-visual alarm: If the mobile terminal is unavailable, the lower part of the meter will continuously emit a high-decibel audio-visual distress signal to attract the attention of people in the surrounding area and obtain rescue.

[0018] The beneficial effects of this invention are: 1. Accurate Judgment: The innovative introduction of "connection status between upper and lower body" as a key dimension for fall judgment, through dual verification of "acceleration + transient characteristics of contact point / magnetic field / other signal quantities", significantly reduces the false alarm rate caused by daily strenuous exercise.

[0019] 2. 24 / 7 monitoring: The split design and dual-battery energy management strategy eliminate the monitoring gap when the device is charging, achieving true 24 / 7 uninterrupted protection.

[0020] 3. Lightweight Algorithm: The core judgment logic is based on clear signal interaction and threshold judgment, without the need for complex calculation models, which reduces the dependence on the computing power of the device's main control chip and helps to reduce costs and power consumption.

[0021] 4. Reliable alarms: The multi-level communication link protection mechanism is designed, with each level progressive from the local machine to the cloud, which greatly improves the probability of successfully sending alarm information in various complex real-world environments.

[0022] 5. Excellent user experience: The split design makes charging and wearing convenient, and the split mode can be automatically and seamlessly switched, reducing user intervention and improving the product's ease of use and acceptance. Attached Figure Description

[0023] Figure 1Here is a flowchart of the fall detection logic; Figure 2 This is a schematic diagram of the connection status signal circuit. Figure 3 This is a schematic diagram of the split structure and limiting mechanism of the wristband. Figure 4 This is a comparison chart of signal characteristic waveforms. Detailed Implementation

[0024] To make the objectives, claims, and advantages of this invention clearer, the present invention will be clearly and completely described below with reference to specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0025] This invention proposes a novel fall detection solution through deep coupling of "mechanical structure + circuit monitoring + judgment algorithm". The core innovation lies in using the physical feature signal threshold of magnetic connection as the fall judgment condition.

[0026] A fall detection system for a wristband based on a magnetically attached split architecture includes a smart wristband, a mobile terminal, and a cloud alarm platform. The smart wristband adopts a split physical architecture, including a detachable upper body 1 and a lower body 2. The upper body 1 houses a first control unit (MCU), a display module, a touch module, and a main battery. The lower body 2 houses a second control unit (MCU), an accelerometer, a backup battery, a wireless communication module, and an audible and visual alarm unit. The second control unit (MCU) is equipped with a connection status monitoring circuit and a voice prompt module.

[0027] The physical connection surfaces of the upper body 1 and the lower body 2 are equipped with a magnetic attraction and conductive mechanism, a mechanical limiting mechanism, and an anti-detachment buckle.

[0028] The magnetic conductive mechanism includes multiple communication contacts 7 (such as spring-loaded connectors) and permanent magnets 6, which are used to achieve electrical connection and physical connection of attraction force between the upper and lower surfaces.

[0029] The permanent magnet 6 includes an upper permanent magnet and a lower permanent magnet, which are respectively embedded in the mating surfaces of the upper and lower surfaces to achieve magnetic adhesion between the upper and lower surfaces.

[0030] The communication contact 7 is a spring-loaded connector that protrudes from the surface of the lower meter body next to the lower permanent magnet. The communication contact is electrically connected to the backup battery and the second control unit MCU of the lower meter body. The communication port 8 is a circular countersunk hole that is embedded in the upper meter body connection surface next to the upper permanent magnet. Its position corresponds one-to-one with the communication contact, so that the communication contact 7 can be inserted into the communication port 8 to achieve electrical connection.

[0031] The mechanical limiting mechanism includes a lower watch body anti-detachment block 5 located at one end of the lower watch body, and an upper watch body with an anti-detachment groove 4 that cooperates with it. The anti-detachment block and the anti-detachment groove are connected by a gap fit to ensure that they can move relative to each other.

[0032] The mechanical limiting mechanism adopts a hinged or snap-on connection design. Its function is to limit the maximum displacement distance of the upper watch body from the lower watch body when it is impacted, so as to prevent the upper watch body from being completely detached and lost. Of course, the mechanical limiting mechanism can be operated manually to completely detach the upper watch body from the lower watch body and remove the upper watch body for charging. At this time, the wearing and detection functions of the lower watch body will not be affected.

[0033] The anti-disengagement buckle includes a button portion 31, a locking tongue body 32, and a return spring 33. The button portion 31 is an arc-shaped protrusion, and the locking tongue body 32 is rigidly connected to the button portion 31. The locking tongue body 32 has a symmetrical double cantilever structure, and cantilever hooks 34 are provided on both sides of the lower end of the locking tongue. The upper watch body has an anti-disengagement buckle mounting cavity for accommodating the anti-disengagement buckle, and the locking tongue body is embedded in the mounting cavity. The cantilever hook portion at the lower end of the locking tongue extends from the opening 9 at the bottom of the upper watch body housing. The upper watch body housing has an opening at the end to allow the button portion to extend. One end of the return spring 34 abuts against the inner wall of the anti-disengagement buckle mounting cavity, and the other end abuts against the button portion 31. The lower watch body connection end has a barb portion 35 corresponding to the cantilever hook 34, and the cantilever hook 34 and the barb portion 35 are hook-shaped and fastened together. The anti-disengagement buckle enables a detachable connection between the upper and lower watch bodies.

[0034] The second control unit MCU of the lower meter body is equipped with a connection status monitoring circuit, which is used to monitor the signal interaction changes of the electrical connection status between the upper meter body and the lower meter body in real time.

[0035] When the bracelet is subjected to a lateral or vertical impact force greater than the magnetic attraction threshold, the magnetic conductive mechanism is disconnected, causing the upper watch body to bounce or shift relative to the lower watch body, resulting in an interruption of the electrical connection.

[0036] The mechanical limiting mechanism, through physical blocking or rotating shaft connection, forces the upper watch body to remain in a non-electrically loose state after the magnetic attraction is disconnected, preventing the upper watch body from flying off and falling off.

[0037] The second control unit (MCU) identifies the following three state characteristics by sampling the signal interaction state or magnetic field / signal quantity changes between the upper and lower gauges: State A (Steady-state connection): Signal interaction is normal, magnetic field strength is stable, or signal quantity is at the reference value, suitable for daily wear or light activity; State B (Physical Separation): Signal interaction is interrupted, magnetic field strength is lower than the separation threshold, or signal quantity undergoes a continuous sudden change, resulting in separation mode or abnormal combination mode of the upper body being removed; State C (Transient Impact): The signal interaction experiences a jump between on and off within a preset time window (T_win), or a sudden change in magnetic field strength / signal quantity followed by recovery, or a return to the connected state after high-frequency oscillation, or a momentary detachment from magnetic attraction caused by a violent impact followed by retraction.

[0038] The wristband fall detection method of the system includes the following steps: Step S1. Dual data stream monitoring: The lower meter body collects data from the accelerometer sensor, and simultaneously collects signal interaction changes between the upper meter body and the upper meter body to scan the electrical connection status. Step S2. Feature fusion determination: Triggering condition: When the accelerometer value is greater than the second threshold, the event determination window is triggered; Step S3 (Multimodal Alarm): After confirming the fall, the system activates the local audible and visual alarm, sends a distress message to the mobile terminal via the wireless communication module, and finally reports it to the cloud alarm platform.

[0039] This system boasts high fall detection accuracy and a significantly reduced false alarm rate. This technical solution innovatively introduces signal connection status as a key dimension for fall detection, constructing a dual detection mechanism based on acceleration signals and transient characteristics of contact / magnetic field / other signal quantities. The separation of magnetic connections requires a specific impact threshold, which is unlikely to be reached by the forces generated during everyday vigorous movements (such as waving, running, or clapping), thus failing to trigger the non-periodic millisecond-level step disconnection characteristic of the contact signal. By fusing these two types of signals, the system can effectively filter out the vast majority of false alarms caused by vigorous movement, significantly improving the reliability of the alarm.

[0040] Furthermore, step S2 also includes a motion interference filtering algorithm based on connection state signal characteristics: 1) When the accelerometer detects a continuous high G-value signal (such as running or clapping), the system analyzes the waveform of the connection status signal; 2) If the connection status signal exhibits periodic, regular microsecond-level oscillations (changes in contact resistance), it is determined to be resonance caused by violent movement, and the fall alarm is disabled; 3) Only when the connection status signal exhibits a non-periodic, step-like millisecond-level disconnection characteristic can it be used as a valid input for fall detection.

[0041] This algorithm is highly efficient and reliable, and requires low computing power from the control unit MCU. This technical solution does not rely on complex neural networks or large-scale sensor data fusion algorithms. Its core determination logic is based on the clear signal interaction and threshold judgment (acceleration G value, contact signal pulse width) of the upper and lower body separation, and the algorithm is simple and efficient. This reduces the computing power and performance requirements of the lower body control unit MCU, allowing the use of lower-cost, lower-power microcontrollers, which helps to reduce hardware costs and extend device battery life.

[0042] Furthermore, the system possesses a seamless dual-battery energy management strategy: 1) Combined Mode: When the upper and lower meters are in a stable connection, the main battery of the upper meter trickle charges the backup battery of the lower meter through the charging pin, while the main battery bears the power consumption of all display and interactive functions. 2) Separation mode: When the upper and lower meter bodies are detected to be separated, the lower meter body immediately switches to backup battery power to ensure continuous monitoring capability during charging or other separation scenarios.

[0043] Therefore, this system provides 24 / 7 uninterrupted fall monitoring, eliminating charging downtime. Traditional integrated wristbands require the user to remove the device while charging, leaving them unprotected during charging (especially during high-risk nighttime hours). This invention's split design and seamless dual-battery energy management completely solve this problem. When the user removes the upper watch body for charging, the lower watch body automatically switches to detached mode, ensuring continuous fall monitoring during charging and achieving true 24 / 7 protection.

[0044] Furthermore, the communication process in step S3 includes a multi-level link protection mechanism: First-level link: The lower table body connects to the mobile terminal APP via Bluetooth BLE, and the alarm is uploaded through the APP's data channel; Secondary link: If the primary link fails (the APP is killed in the background), the table below switches to Bluetooth broadcast mode. After the background service at the mobile terminal operating system level (such as Android's accessibility service or iOS's background process) hears the broadcast, it can automatically wake up or start the alarm application. Level 3 Link: If the mobile terminal is unavailable, the lower body continuously emits a high-decibel sound and light distress signal to attract the attention of the surrounding crowd.

[0045] The alarm link of this system is reliable, ensuring that distress messages are delivered. To address real-world issues such as potential system termination of mobile apps and poor network connectivity, this system employs a multi-level link protection mechanism. Prioritizing communication via a wireless communication module (Bluetooth is preferred in this example), the system connects to the app for reporting. If this fails, Bluetooth broadcasting and system-level background services (such as Android's accessibility services or iOS's background processes) are used to wake up the app or launch the alarm application. Finally, when the mobile terminal is completely unavailable, a local high-decibel audible and visual alarm is activated. This progressive communication strategy significantly increases the probability of successful alarm activation in complex real-world environments, providing users with multiple layers of security.

[0046] Furthermore, the magnetic attraction force (F_mag) of the magnetic attraction conductive mechanism is specifically calibrated: Set the threshold range of F_mag to be greater than the centrifugal force generated by a daily hand wave (F_daily), but less than the instantaneous shear force generated when the wrist hits the ground when the human body falls (F_fall), to ensure that the upper and lower body limits are disengaged only under the impact of a fall.

[0047] 1. The system structure and circuit design are as follows: (See attached manual) Figure 1 , 2 .

[0048] 1.1 Split Architecture: The lower part (Host): Worn close to the wrist, it houses a second MCU, an accelerometer, a backup battery (e.g., 60mAh), a wireless communication module, a connection status monitoring circuit, and an audible and visual alarm. Its primary function is to monitor vital signs and fall detection.

[0049] The upper display panel (Display): Contains the first MCU, display screen, touch screen, and main battery (large capacity). It is primarily responsible for information display, user interaction, and powering the main system.

[0050] 1.2 Connection Interface (e.g.) Figure 1 ): The upper meter body charges the lower meter body through the 7 charging ports; In the structure, the upper table body 1 and the lower table body 2 communicate through communication port 6; Four magnetic attraction and conductive mechanisms are arranged on one side, which are responsible for electrical connection and main attraction force.

[0051] It is equipped with 3 anti-detachment buckles and 5 limit mechanisms as physical safety to prevent the upper body from flying off completely after the magnetic attraction is disconnected.

[0052] 1.3 Detection Circuit: The lower body control unit MCU is a connection signal interaction monitoring node. This node can be any type of communication contact, GPIO level detection point, Hall sensor output, or other signal sensing terminal. When the connection is stable, a stable signal is detected; when an impact causes displacement or disconnection, the signal changes abruptly or is interrupted. The connection status monitoring circuit is configured to monitor at least one of the following signals to generate a physical characteristic signal of the conductive connection between the upper and lower body: (a) The electrical on or off state of a communication contact, regardless of whether the contact carries serial communication, parallel communication, I2C, SPI protocol or a simple level signal; (b) Changes in the magnetic field strength output by a magnetic field sensor located inside the body of the instrument; (c) Changes in the mutual inductance or capacitive coupling signal between the coils of the meter body.

[0053] 2. For the multi-level fall detection algorithm flowchart, please refer to the instruction manual appendix. Figure 3 and attached Figure 4 .

[0054] S1 Data Acquisition: Data 1, Accelerometer (50Hz sampling), such as setting the first threshold of the accelerometer: acceleration is 5G, the second threshold: acceleration is 3G; Data 2, Monitoring of electrical connection status signal of upper and lower meter body (interrupt trigger or 100Hz polling), when the acceleration is >3G, the upper and lower meter body will separate. S2 Impact Detection: When the acceleration > 3G, the signal fusion judgment process begins; S3 signal fusion is a core technology; please refer to the appendix. Figure 4 : Scenario A (Suspected fall or minor fall): Acceleration > 3G, and simultaneously, the electrical connection status signal between the upper and lower watch bodies undergoes a single "on→off→on" transition within a preset time, for example, a pulse width of 10ms-500ms. At this time, a voice prompt will ask if an alarm needs to be triggered; the alarm can be canceled by using the function button on the wristband.

[0055] Judgment: If no choice is made, the system will default to triggering an alarm signal after 3 seconds if the person has fallen.

[0056] Scenario B (Severe Drop): Acceleration > 5G.

[0057] Judgment: Issue an alarm signal directly.

[0058] Scenario C (vigorous movement / clapping): Acceleration > 3G and is periodic, while the electrical connection status signal of the upper and lower body shows a cyclical "on → off → on" jump or high-frequency glitches (microsecond-level oscillations).

[0059] Judgment: Ignore.

[0060] 3. Mode Switching Remove the upper table body: the lower table body enters the separation mode.

[0061] Adsorb the upper surface body: Enter the combined mode.

[0062] 4. Alarm Communication Process Once a fall is confirmed: Contact signal layer: Immediately releases any mute settings on the upper body.

[0063] Local layer: The lower body emits a high-decibel alarm sound and strong vibration.

[0064] Link layer: The following table body sends emergency commands to the mobile phone via Bluetooth (BLE 5.0).

[0065] If the mobile app is active: The app directly uploads GPS location and alarm information to the cloud.

[0066] If the app is killed in the background: try to wake it up using Android accessibility services or iOS system-level notification permissions, or trigger a keep-alive mechanism via Bluetooth broadcast (Beacon).

[0067] Cloud layer: The cloud platform receives alarms, sends APP push notifications and SMS messages to children, and automatically dials emergency contacts.

[0068] 5. Working principle of this system: The system's preset magnetic attraction is sufficient to resist the swaying force of everyday movements (<2G). However, when a fall occurs (typically >3G~5.5G impact), the shear force generated by the wrist hitting the ground or the body vibrating violently will instantly overcome the magnetic attraction, causing the upper body to bounce or shift slightly under the constraint of the mechanical limit (hook). This displacement can cause the 4-pin contact to momentarily disconnect (tens of milliseconds) and then be pulled back by the magnetic force, or become a poor contact; or cause a step change in the magnetic field strength detected by the internal Hall sensor, or cause a sudden change in the signals of other sensors.

[0069] The second control unit (MCU) simultaneously captures both the high-G impact signal and the momentary circuit interruption signal, thus constituting a dual determination of a fall. This signal interaction phenomenon is extremely difficult to reproduce when clapping or waving (because there is no displacement of the upper and lower body due to impact), thereby greatly reducing false alarms.

[0070] This magnetically attached, split-type wristband fall detection system consists of an upper and lower watch body, connected by magnetic contacts and a mechanical limiting structure. The lower watch body is responsible for collecting data from the accelerometer and connection status signals, which are then fused and determined by the second control unit (MCU). When a fall is detected, the system activates a local alarm and sends a command to the mobile terminal via Bluetooth, ultimately uploading the data to a cloud platform. The system can continuously monitor when the wristband is detached.

[0071] Meanwhile, the split design not only serves the core detection function but also brings convenience to use. The upper part can be removed separately for charging or replacement, while the lower part can still be easily worn and used for monitoring. The seamless switching between the split mode and the automatic charging when combined reduce manual intervention by the user, improving the product's ease of use and user acceptance.

[0072] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the appended claims.

Claims

1. A fall detection system for a smart bracelet based on a magnetically attached split architecture, comprising a smart bracelet, a mobile terminal, and a cloud-based alarm platform; characterized in that, The smart bracelet adopts a split upper and lower body architecture, including a separable upper body and a lower body. The physical connection surfaces of the upper and lower bodies are respectively equipped with magnetic conductive mechanisms and movable connectors, realizing magnetic conductivity and separation limiting between the upper and lower bodies. The upper body houses a first control unit (MCU), a display module, a touch module, and a main battery. The lower body houses a second control unit (MCU), an accelerometer, a backup battery, a wireless communication module, and an audible and visual alarm unit. The second control unit (MCU) of the lower body is equipped with a connection status monitoring circuit and a voice prompt module, used to monitor the physical characteristic signals of the conductive connection between the upper and lower bodies in real time. The lower body collects accelerometer and connection status signal data, which are then fused and determined by the second control unit (MCU). The second control unit (MCU) simultaneously captures two signals: a high-G impact and a circuit interruption, forming a dual fall signal. When a fall is detected, the system activates a local alarm and sends a command to the mobile terminal via Bluetooth, ultimately uploading the data to the cloud platform. The upper body can be detached and completely separated from the lower body, while the lower body can continue to perform monitoring.

2. The wristband fall detection system based on a magnetic split architecture according to claim 1, characterized in that, The magnetic attraction conductive mechanism includes multiple communication contacts and a permanent magnet, which are used to provide electrical and physical adsorption connections between the upper and lower watch bodies. The movable connector includes a mechanical limiting mechanism and an anti-detachment buckle. The mechanical limiting mechanism is located on the side of the watch body closer to the magnetic attraction conductive mechanism, forming a hinged or snap-fit ​​connection, which limits the maximum displacement distance of the upper watch body from the lower watch body. The anti-detachment buckle is located on the opposite side, realizing a detachable connection between the upper and lower watch bodies.

3. The wristband fall detection system based on a magnetic split architecture according to claim 1, characterized in that, The connection status monitoring circuit is configured to monitor at least one of the following signals to generate a physical characteristic signal for the conductive connection of the upper and lower body: (a) The electrical on or off state of a communication contact, regardless of whether the contact carries serial communication, parallel communication, I2C, SPI protocol or a simple level signal; (b) Changes in the magnetic field strength output by a magnetic field sensor located inside the body of the instrument; (c) Changes in the mutual inductance or capacitive coupling signal between the coils of the meter body; wherein the signal interaction change is caused by the relative displacement or transient separation of the upper and lower meter bodies due to external force impact.

4. The wristband fall detection system based on a magnetic split architecture according to claim 1, characterized in that, The mechanical limiting mechanism is configured as a semi-separated protection structure. When the wristband is subjected to an impact force greater than the magnetic attraction threshold, the magnetic attraction conductive mechanism is disconnected, and the mechanical limiting mechanism forces the upper body to remain in a non-electrically loose state near the lower body.

5. A wristband fall detection system based on a magnetic split architecture according to claim 1, characterized in that, The physical characteristic signal used for real-time monitoring of the conductive connection between the upper and lower body has three states: Steady-state connection: Signals interact normally, magnetic field strength is stable, or signal quantity is at the reference value; Physical separation: signal interruption, magnetic field strength below the separation threshold, or continuous abrupt change in signal quantity; Transient impact: The signal undergoes a 'on→off→on' transition within a preset time window, or the magnetic field strength / signal quantity undergoes a sudden change and then recovers.

6. A method for detecting wristband falls based on the system according to any one of claims 1-6, characterized in that, Includes the following steps: S1. Dual data stream monitoring: The lower body collects data from the accelerometer sensor, and at the same time collects and scans the signal interaction changes of the electrical connection status with the upper body; S2. Feature Fusion Judgment: When the collected combined acceleration sensor value exceeds the first preset threshold, the event judgment window is entered, and the fall judgment is made by combining the physical feature signal of the electrical connection between the upper and lower body. S3. Multimodal alarm: After confirming a fall, the system activates a local audible and visual alarm and sends a distress message to the mobile terminal via a wireless communication module, and finally reports it to the cloud alarm platform.

7. The method for detecting a wristband fall according to claim 7, characterized in that, The feature fusion determination step includes: (1) Suspected fall or minor fall: Acceleration > 3G, and at the same time, the electrical connection status signal of the upper and lower body of the watch body shows a single "on → off → on" jump within a preset time, such as a pulse width of 10ms-500ms; at this time, a voice prompt asks whether to alarm; at this time, the alarm can be canceled by using the function button set on the wristband. Judgment: If no choice is made, the system will default to triggering an alarm signal after 3 seconds if the person has fallen. (2) Heavy fall: Acceleration > 5G, judgment: directly issue an alarm signal; (3) Vigorous exercise / clapping: acceleration > 3G and is periodic, while the electrical connection status signal of the upper and lower body shows a cyclic "on → off → on" jump or high frequency glitch; Judgment: Ignore.

8. The method for detecting a wristband fall according to claim 7, characterized in that, It also includes a motion interference filtering algorithm, which distinguishes between vigorous motion and a real fall by analyzing the connection state signal waveform. Specifically: S1. When the accelerometer detects a continuous high G-value signal, the system analyzes the waveform of the connection status signal; S2. If the connection status signal exhibits periodic, regular microsecond-level high-frequency oscillations, it is determined to be mechanical resonance caused by strenuous exercise (such as running or clapping), and the system disables the fall alarm; S3. Only when the connection status signal exhibits a non-periodic, step-like millisecond-level disconnection characteristic can it be used as a valid input for fall detection.

9. A method for detecting a wristband fall according to claim 7, characterized in that, The alarm process includes a multi-level link protection mechanism, including Bluetooth connection, system-level background service monitoring, and local audible and visual alarms; specifically: Bluetooth connection: The lower part of the meter connects to the mobile terminal APP via Bluetooth BLE, and the alarm is uploaded through the APP's data channel; Background service monitoring: If the primary link fails and the APP is detected and killed in the background, the table below switches to Bluetooth broadcast mode. After the system-level background service of the mobile terminal (such as the accessibility service of Android or the background process of iOS) hears the specific broadcast, it automatically wakes up the APP process and performs alarm uploading. Local audio-visual alarm: If the mobile terminal is unavailable, the lower part of the meter will continuously emit a high-decibel audio-visual distress signal to attract the attention of people in the surrounding area and obtain rescue.

Images