A wearable labor worker wide-area non-inductive attendance and real-name positioning management system

Abstract

The embodiment of the application provides a wearable labor worker wide-area non-inductive attendance and real-name positioning management system, a pulse wave sensor of a smart bracelet terminal is used for collecting a photoplethysmogram (PPG) signal of a worker in real time, a local microcontroller is used for running a deep learning model, and an unforgeable pulse wave biological feature vector is extracted from the PPG signal; the feature vector is compared with a worker identity template pre-stored in the bracelet immediately, and only when the comparison is successful, the microcontroller associates identity information of the current worker with real-time position coordinates and a time stamp acquired by a satellite positioning module, and generates an effective attendance record. Subsequently, the record is transmitted to a communication base station in a low-power wide-area mode through a LoRa communication module, and is forwarded to a cloud management platform by the base station, and a complete real-name positioning and attendance process is completed.

Description

Technical Field

[0001] This invention relates to the field of smart wearable technology, and in particular to a wearable, wide-area, contactless attendance and real-name location management system for laborers. Background Technology

[0002] As projects such as new energy power plant construction expand to remote areas such as deserts and wastelands, the wide construction areas and dispersed personnel pose severe challenges to labor management. Currently, the closest existing technology to this solution is smart wearable devices (such as smart work badges and positioning wristbands) that combine satellite positioning and wireless communication. These devices typically manage workers by acquiring their location through a built-in satellite positioning module, using electronic fences for attendance tracking and area control, and transmitting the location data back to the management platform via cellular networks (such as 4G / 5G) or other wireless methods.

[0003] However, when faced with the vast, network-free, and highly mobile construction environments of remote areas such as deserts, the aforementioned existing technologies suffer from the following inherent defects and limitations, causing serious negative impacts on practical applications. First, the identity verification mechanism of existing equipment has fundamental loopholes. Fingerprint and facial recognition methods rely on sensors on the device's surface, making them easily deceived by non-living means such as photos, videos, or bionic fingerprints, failing to ensure that the device user and the registered worker are the same person. This technical defect directly leads to frequent instances of "proxy clocking in," severely distorted attendance records, resulting in labor disputes and creating management loopholes where contractors can falsely report worker numbers and embezzle labor costs. Second, there is an irreconcilable contradiction between the device's battery life and the need for continuous positioning. To achieve meter-level accuracy, existing equipment requires high-frequency activation of satellite positioning modules with power consumption reaching tens of milliamps. However, limited by the size of wearable devices, their battery capacity is typically less than 500 mAh, resulting in continuous positioning battery life of less than 12 hours. This technical flaw forces equipment to frequently stop working to recharge in scenarios requiring workers to operate continuously in the field for several days, resulting in a large amount of missing critical positioning data and rendering the electronic fence and trajectory playback functions ineffective. Third, the equipment's anti-tampering mechanism has a structural failure risk. Most existing smart bracelets use simple physical clips or screws for fixing, lacking an integrated design that combines effective electronic detection with physical protection. They can be disassembled in seconds using simple tools. This technical flaw directly leads to the "human-machine separation" problem; once the equipment is disassembled, all positioning and attendance data lose their binding relationship with the actual personnel, making it impossible to prevent workers from disassembling the equipment to evade supervision.

[0004] In summary, existing technical solutions have significant shortcomings in terms of identity verification, ultra-long-range positioning, and anti-tampering and anti-detachment, and cannot meet the requirements of refined management of laborers at large construction sites in remote areas such as deserts and wastelands. Summary of the Invention

[0005] This invention provides a wearable, wide-area, contactless attendance and real-name location management system for laborers, which solves the technical problems in existing attendance management systems, such as easy identity theft and proxy clocking, contradiction between short battery life of positioning devices and the need for continuous positioning, failure of anti-tampering mechanisms leading to separation of human and machine, and data loss and unreliability in the absence of network.

[0006] In a first aspect, embodiments of the present invention provide a wearable, wide-area, contactless attendance and real-name location management system for laborers, including a smart bracelet terminal, a communication base station, and a cloud management platform; The smart bracelet terminal includes: a pulse wave sensor for attaching to a worker's skin to collect photoplethysmography (PPG) signals, a satellite positioning module for acquiring real-time location coordinates, a LoRa communication module for communicating with a communication base station, and a microcontroller electrically connected to the pulse wave sensor, the satellite positioning module, and the LoRa communication module, respectively. The communication base station is communicatively connected to both the smart bracelet terminal and the cloud management platform. The microcontroller is configured to: The PPG signal is received, and a deep learning model is run to extract the pulse wave biofeature vector from the PPG signal; The extracted pulse wave biometric vector is matched with a pre-stored worker identity template; If a match is successful, the worker's identity will be associated with the real-time location coordinates and / or current timestamp from the satellite positioning module to generate a valid attendance record. The valid attendance records are sent to the communication base station via the LoRa communication module, and then forwarded to the cloud management platform by the communication base station.

[0007] Preferably, the smart bracelet terminal also includes a wristband and multiple anti-tamper mechanisms disposed on the wristband; The multiple anti-tamper mechanism includes: a continuous metal wire loop embedded inside the wristband, a pressure sensor array arranged on the skin contact surface of the wristband, and an optical proximity sensor disposed at the buckle of the wristband. The microcontroller is electrically connected to the continuous metal wire loop, the pressure sensing array, and the optical proximity sensor. When the microcontroller detects that the continuous metal wire loop is disconnected, the pressure distribution pattern collected by the pressure sensing array deviates from the preset threshold range, or the optical proximity sensor detects that the latch is opened, it triggers an anti-tamper alarm, sends the anti-tamper alarm to the cloud management platform through the LoRa communication module, and locks the smart bracelet terminal.

[0008] Preferably, the multiple anti-tamper mechanism also includes a cut-resistant wristband body made of Kevlar fiber braided layer, a metal buckle with a double locking mechanism, and an ultra-high frequency radio frequency identification tag embedded inside the wristband; the metal buckle is provided with an asymmetrical keyhole that matches a special irregular key.

[0009] Preferably, the smart bracelet terminal has a built-in time-sharing positioning state machine, which is configured to include at least the following states: In high-speed positioning mode, the satellite positioning module operates continuously for a first positioning cycle. In a low-power positioning state, the satellite positioning module periodically wakes up and performs positioning at a second positioning period that is longer than the first positioning period. In deep sleep mode, the satellite positioning module is completely shut down; The time-division positioning state machine switches between high-speed positioning state, low-power positioning state, and deep sleep state according to the current working scenario of the smart bracelet terminal.

[0010] Preferably, the time-division positioning state machine also includes an abnormal alarm state; When an anti-tamper alarm, distress signal, or electronic fence boundary crossing event is triggered, the time-division positioning state machine switches to the abnormal alarm state; in the abnormal alarm state, the satellite positioning module operates continuously for a third positioning cycle that is less than the first positioning cycle, and the LoRa communication module sends alarm information to the communication base station at maximum transmission power.

[0011] Preferably, the cloud management platform or the smart bracelet terminal stores three-level electronic fence configuration information, wherein the three-level electronic fence includes a safe zone, a buffer zone, and a restricted zone. The microcontroller of the smart bracelet terminal is configured as follows: Obtain the real-time position coordinates from the satellite positioning module; The type of area where the worker is located is determined based on the real-time location coordinates and the configuration information of the three-level electronic fence. When it is determined that a worker has entered the buffer zone, an early warning message is generated; When it is determined that a worker has entered the restricted area, an emergency alarm is generated, and an audio-visual vibration alarm is output through the audio-visual vibration module of the smart bracelet terminal.

[0012] Preferably, the smart bracelet terminal further includes an offline cache module, which is electrically connected to the microcontroller; The microcontroller is configured to store the generated attendance records and location data in the offline cache module when a communication interruption with the communication base station is detected. The offline caching module adopts a circular overwrite storage strategy, and the storage capacity is configured to cache data for at least 72 hours. The microcontroller is also configured to use an SM3 hash chain structure to protect the integrity of the stored data. The hash chain is constructed by using a random number generated when the smart bracelet terminal is first activated as the initial hash value, concatenating the current hash value with the current data record, and then performing an SM3 operation to obtain the next hash value.

[0013] Preferably, the LoRa communication module communicates with the communication base station using a time-division multiple access protocol; The communication base station or the smart bracelet terminal is configured with frame structure parameters, which divide time into communication frames of equal length, and each communication frame contains multiple time slots. The microcontroller is configured to control the LoRa communication module to send data within the time slot allocated to the smart bracelet terminal according to a preset time slot allocation rule.

[0014] Preferably, the deep learning model is an improved Inception-Time convolutional neural network; The improved Inception-Time convolutional neural network includes: multiple Inception modules connected in series, and a global average pooling layer and a fully connected layer connected after the last Inception module; Each Inception module includes four parallel branches: a 1×1 convolutional layer, a 3×3 convolutional layer, a 5×5 convolutional layer, and a 3×3 max pooling layer. The outputs of the four parallel branches are fused by a channel splicing layer to output multi-scale features. The fully connected layer outputs a 128-dimensional feature vector, which serves as the biofeature vector of the pulse wave.

[0015] Preferably, the satellite positioning module has dual-mode positioning functionality of the Global Positioning System (GPS) and the BeiDou Navigation Satellite System (BDS), and is configured to receive and process satellite signals from the GPS and / or BDS.

[0016] This invention provides a wearable, wide-area, contactless attendance and real-name location management system for laborers. It abandons traditional attendance methods that rely on external features (such as fingerprints and faces), instead utilizing the unique, intrinsic pulse wave biometrics of the human body as a liveness anchor for identity authentication. The pulse wave sensor on the smart bracelet terminal collects the worker's photoplethysmography (PPG) signal in real time, and a local microcontroller runs a deep learning model to extract an unforgeable pulse wave biometric vector from the PPG signal. This vector is then compared with a worker identity template pre-stored in the bracelet. Only when a match is successful does the microcontroller associate the current worker's identity information with the real-time location coordinates and timestamp obtained by the satellite positioning module to generate a valid attendance record. Subsequently, this record is transmitted to a communication base station via a LoRa communication module in a low-power, wide-area manner, and then forwarded by the base station to the cloud management platform, completing a full real-name location and attendance process. Compared with existing technologies, this system offers the following advantages: (1) Since pulse wave signals have the properties of living bodies and are difficult to be copied by non-living means (such as photos, videos or bionic fingerprints), the phenomenon of proxy attendance and identity theft is completely eliminated from the technical principle, and the authenticity of attendance records is fundamentally guaranteed.

[0017] (2) It achieves true contactless attendance. Workers do not need to actively perform any clocking-in action. The system automatically completes identity verification and record generation in the background, without affecting normal construction operations.

[0018] (3) By localized feature extraction and matching, the reliance on real-time network transmission is reduced. Combined with the long-distance and low-power characteristics of LoRa communication, the system can still operate stably in remote areas without cellular network coverage, such as deserts and wastelands. It can also reliably transmit the real-name and location data of the bound personnel back to the management platform, thus providing a precise, reliable and efficient solution for real-name management and attendance of laborers for large-scale field construction projects. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a structural block diagram of the wearable worker wide-area contactless attendance and real-name positioning management system provided in an embodiment of the present invention; Figure 2 This is a flowchart of the identity binding process provided in an embodiment of the present invention; Figure 3This is a schematic diagram of the composition of a 10-dimensional feature vector provided in an embodiment of the present invention; Figure 4 The diagram shows the time-division positioning state machine provided in the embodiment of the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0022] Figure 1 This is a wearable, contactless attendance and real-name location management system for laborers according to an embodiment of the present invention, with reference to... Figure 1 The system includes a smart bracelet terminal 100, a communication base station 200, and a cloud management platform 300.

[0023] The smart bracelet terminal 100 includes: a pulse wave sensor 110 for attaching to a worker's skin to collect photoplethysmography (PPG) signals, a satellite positioning module 140 for acquiring real-time location coordinates, a LoRa communication module 130 for communicating with a communication base station 200, and a microcontroller 120 that is electrically connected to the pulse wave sensor 110, the satellite positioning module 140, and the LoRa communication module 130, respectively.

[0024] The PPG signal (Photoplethysmography signal) is an optical signal used to detect changes in intravascular blood volume. In this embodiment, the pulse wave sensor 110 at the bottom of the wristband collects optical signals at two wavelengths, 560nm and 940nm, at a sampling rate of 100Hz. These signals reflect changes in microvascular blood flow caused by heartbeats. The 560nm wavelength light is mainly absorbed by oxyhemoglobin, making it sensitive to changes in microvascular blood flow; the 940nm wavelength light has a greater penetration depth, allowing for the acquisition of richer tissue information. The dual-wavelength signals are bandpass filtered (passband range of 0.5 to 10 Hz) to remove baseline drift and high-frequency noise. The deep learning model refers to an improved Inception-Time convolutional neural network, which can automatically learn and extract highly discriminative features from the raw PPG signal. The pulse wave biometric vector is a 128-dimensional numerical vector output by this deep learning model, serving as a unique biometric identifier for an individual. After a worker puts on the wristband, sensors continuously attach to the skin to collect PPG signals. The microcontroller 120 inside the wristband immediately runs the deep learning model, converting the raw physiological signals into quantifiable feature vectors. Existing fingerprint or facial recognition technologies are easily fooled by non-living entities, while this step utilizes the liveness attribute of pulse waves (only a living person's heartbeat can generate regular PPG signals), eliminating forgery methods such as photos and videos at the signal source, and providing an unforgeable foundation for achieving genuine identity authentication.

[0025] The communication base station 200 is communicatively connected to the smart bracelet terminal 100 and the cloud management platform 300, respectively.

[0026] The microcontroller 120 is configured to: The PPG signal is received, and a deep learning model is run to extract the pulse wave biofeature vector from the PPG signal.

[0027] The extracted pulse wave biometric vector is matched with a pre-stored worker identity template.

[0028] If a match is successful, the worker's identity will be associated with the real-time location coordinates and / or current timestamp from the satellite positioning module 140 to generate a valid attendance record.

[0029] In this system, worker identity refers to a unique identifier (such as a worker ID) pre-stored in the wristband. The satellite positioning module 140 supports GPS / BeiDou dual-mode positioning, providing the latitude and longitude coordinates of the wristband's current location. The timestamp is provided by the system's internal real-time clock. A valid attendance record is a structured data entry containing worker identity, time, and location information, and this record is only recognized after identity verification. In construction scenarios, workers are scattered across work areas of several square kilometers, and traditional clocking-in methods cannot prove actual attendance. This embodiment of the invention only binds worker identity with current location and time after successful identity matching, meaning that workers must be wearing the wristband and pass liveness verification for their location and time information to be recorded as valid attendance. In this way, each attendance record simultaneously provides multiple pieces of evidence: "who, when, where," and "the person is wearing it." Any situation where the wristband is not worn or is held by someone else cannot generate a valid record, thus ensuring the authenticity and non-repudiation of attendance data.

[0030] The valid attendance record is sent to the communication base station 200 via the LoRa communication module 130, and then forwarded to the cloud management platform 300 by the communication base station 200.

[0031] Specifically, LoRa (Long Range, a low-power wireless communication technology) is a low-power wide-area network communication technology characterized by long range and low power consumption. In this embodiment of the invention, the LoRa communication module 130 and the communication base station 200 use spread spectrum communication, achieving a coverage radius of up to 15 kilometers in suburban areas under line-of-sight conditions. The communication base station 200 is deployed at a high point on the construction site, integrating a LoRa gateway and a 4G / 5G module, responsible for receiving wristband data and forwarding it to the cloud via cellular network or fiber optic cable. The cloud management platform 300 is deployed in the enterprise data center, providing functions such as personnel information management, attendance statistics, and location monitoring. In remote areas such as deserts, cellular network coverage is sparse, and traditional attendance devices often cannot transmit data due to lack of signal, resulting in data loss.

[0032] This invention leverages the LoRa communication feature, which does not rely on public cellular networks, to build a self-contained, wide-coverage local network. This allows the wristband to send attendance records to a local base station even in environments without 4G / 5G signals. The base station then uses its high-bandwidth connection (solar-powered 4G / 5G or fiber optic) to forward the data to the cloud. This not only solves the data backhaul problem in wide-area, network-free environments but also allows the wristband to last for over 30 days due to LoRa's ultra-low power consumption (average operating current of 1.2mA), meeting the needs of continuous fieldwork. Ultimately, the cloud platform aggregates the real-name attendance and location data of all workers, providing managers with real-time, reliable personnel distribution and attendance statistics.

[0033] Based on the above embodiments, as a preferred implementation, the smart bracelet terminal 100 further includes a wristband and multiple anti-tamper mechanisms disposed on the wristband.

[0034] The multiple anti-tamper mechanism includes: a continuous metal wire loop 152 embedded inside the wristband, a pressure sensor array 153 arranged on the surface of the wristband in contact with the skin, and an optical proximity sensor 151 disposed at the buckle of the wristband.

[0035] Based on the aforementioned structure, the multi-layer anti-tamper mechanism further includes a cut-resistant wristband body made of Kevlar fiber braided layers, a metal buckle with a double locking mechanism, and an ultra-high frequency radio frequency identification (RFID) tag 154 embedded inside the wristband; the metal buckle has an asymmetrical keyhole that matches a special irregularly shaped key. Specifically, the wristband body adopts a multi-layer composite structure: the outermost layer is a high-strength polyurethane coating, the middle layer is a Kevlar fiber braided layer with a flexible circuit layer embedded within it, and the innermost layer is a skin-friendly silicone layer. The tensile strength of the Kevlar fiber reaches 3.5 gigapascals (GPa), the wristband width is 25 mm, the thickness is 3.2 mm, and it can withstand a tensile force of 200 Newtons without breaking. A buckle is provided at the end of the wristband to fix it in a closed loop shape.

[0036] Furthermore, a 0.3 mm diameter stainless steel wire is laid along the length of the wristband between the Kevlar fiber braided layer and the flexible circuit layer. This stainless steel wire is connected at both ends of the wristband via pads on the flexible circuit board, forming a continuous conductive loop. The nominal resistance of this loop is 10 ohms (Ω), with a tolerance of ±2Ω. To accurately measure the loop resistance and eliminate the influence of contact resistance, a four-wire measurement method is used: the microcontroller 120 applies a constant excitation current (e.g., 1 mA) to the loop through two independent current leads, while simultaneously measuring the voltage drop across the loop directly through two other independent voltage detection leads. According to Ohm's law, the loop resistance R = V / I, where V is the detected voltage drop and I is the applied excitation current. Since almost no current flows through the voltage detection loop, the contact resistance of the wires themselves does not introduce measurement errors. The microcontroller 120 performs this resistance measurement cyclically at a frequency of 10 times per second. When the wristband is cut by scissors or a knife, the stainless steel wire will inevitably break, causing the circuit resistance to jump instantly from about 10Ω to infinity (an actual measured value exceeding 10 kilohms is considered a break).

[0037] Furthermore, 16 miniature pressure sensors are evenly arranged along the circumference of the wristband on the inner surface of the skin-friendly silicone layer in contact with the skin. The center-to-center spacing between adjacent sensors is 15 mm. Each pressure sensor has a detection range of 0 to 50 kPa and outputs an analog voltage signal proportional to the applied pressure. This signal is converted into a digital value by an analog-to-digital converter (ADC) and then sent to the microcontroller 120. Under normal wearing conditions, the wristband maintains a moderate fit with the worker's wrist skin, and the output values ​​of the 16 pressure sensors form a stable spatial distribution pattern. For example, the pressure value on the back of the wrist (corresponding to sensor numbers 1-4) is typically 10-20 kPa, and the pressure value on the palm side of the wrist (corresponding to sensor numbers 9-12) is 5-15 kPa. This distribution pattern is pre-collected and stored as a preset threshold range (i.e., the normal pressure range allowed for each sensor, and the allowable range of pressure difference between adjacent sensors). When a worker attempts to remove the wristband from their wrist, the contact between the wristband and skin changes drastically: the pressure of some sensors drops sharply (e.g., below 1 kPa), the pressure of others returns to zero, and the original pressure gradient distribution is disrupted. The microcontroller 120 compares the 16 pressure values ​​collected in real time with preset threshold ranges one by one. When the deviation exceeds a preset proportion (e.g., the pressure values ​​of more than 4 sensors are simultaneously below 50% of the normal lower limit), it is determined that the pressure distribution pattern is abnormal.

[0038] Furthermore, an infrared optical proximity sensor 151 is installed inside the metal buckle of the wristband. This sensor integrates an infrared light-emitting diode (LED) and a photodetector. When the buckle is closed, the reflection distance between the buckle's tongue and the sensor is less than 1 mm, the photodetector receives a high intensity of reflected light, and the sensor outputs a high-level signal (e.g., logic "1"). When the buckle is opened, the tongue moves away from the sensor, the intensity of reflected light drops sharply, and the sensor output transitions to a low-level signal (logic "0"). The sensor's output is directly connected to a digital input pin of a microcontroller 120, which reads the pin level in millisecond cycles.

[0039] The microcontroller 120 is electrically connected to the continuous metal wire loop 152, the pressure sensor array 153, and the optical proximity sensor 151. When the microcontroller 120 detects that the continuous metal wire loop 152 is disconnected, the pressure distribution pattern collected by the pressure sensor array 153 deviates from the preset threshold range, or the optical proximity sensor 151 detects that the latch is opened, it triggers an anti-tamper alarm, sends the anti-tamper alarm to the cloud management platform 300 through the LoRa communication module 130, and locks the smart bracelet terminal 100.

[0040] The microcontroller 120 is electrically connected to the measurement circuit of the continuous metal wire loop 152 (via a four-wire interface), the 16 analog-to-digital conversion channels of the pressure sensor array 153, and the digital output pin of the optical proximity sensor 151. The microcontroller 120 internally has a continuously running tamper detection task that cyclically performs the following operations at a frequency of at least 100 Hz: Read the circuit resistance measurement value and determine if it is within the nominal range of 8Ω to 12Ω. If the resistance value is greater than 12Ω and the duration exceeds 0.5 seconds, the wire circuit is considered broken.

[0041] The current values ​​of 16 pressure sensors are read and compared with a preset pressure distribution pattern stored in the non-volatile memory of the microcontroller 120. If the pressure values ​​of at least 4 sensors are simultaneously below 50% of their normal lower limit, or if the pressure values ​​of all 16 sensors are below 1 kPa, the pressure distribution pattern is determined to have deviated from the preset threshold range.

[0042] Read the output level of the optical proximity sensor 151. If the level changes from high to low and remains stable at a low level for more than 0.2 seconds, the latch is determined to be open.

[0043] When any of the above conditions are met, the microcontroller 120 immediately triggers an anti-tamper alarm. Specifically, the microcontroller 120 performs the following operations: First, generate an alarm data packet. The alarm data packet must contain at least the following fields: alarm type (encoded in one byte, for example, 0x01 indicates a broken wire loop, 0x02 indicates abnormal pressure distribution, and 0x03 indicates a latch is open), alarm trigger time (read from the real-time clock circuit, accurate to the second), current satellite positioning coordinates (the latest latitude and longitude values ​​read from the satellite positioning module 140), and the device's unique serial number.

[0044] Second, an alarm is sent via the LoRa communication module 130. The microcontroller 120 sends the alarm data packet to the transmission queue of the LoRa communication module 130 and sets it to the highest priority. Regardless of the current operating state of the LoRa communication module 130 (sleep, waiting to transmit, or transmitting), the microcontroller 120 forcibly wakes it up and configures the transmission power to the maximum power (20dBm), using the minimum spreading factor SF7 to ensure the fastest transmission rate. The alarm data packet is sent to the communication base station 200 deployed at the construction site via the LoRa wireless link. The communication base station 200 then forwards the alarm data packet to the cloud management platform 300 in real time via a 4G / 5G cellular network or fiber optic cable. After receiving the alarm, the cloud management platform 300 highlights the location of the device on an electronic map and pushes an alarm notification to the mobile application of relevant management personnel.

[0045] Third, the attendance and location functions of the smart bracelet terminal 100 are locked. The microcontroller 120 internally sets a status bit called "Lock Flag" and sets it from 0 to 1. After the lock flag is set, the system executes the following prohibition logic: When step three (generation of attendance records after successful identity matching) is triggered, the microcontroller 120 first checks the lock flag. If the lock flag is 1, the attendance record generation step is skipped directly, and no new valid attendance records are generated. When step four (uploading location data) is executed periodically, the microcontroller 120 also checks the lock flag. If the lock flag is 1, it stops obtaining location coordinates from the satellite positioning module 140 and does not send any regular location data through the LoRa communication module 130. It should be noted that the lock flag does not affect the transmission of the tamper alarm itself—even if the bracelet is in locked mode, the microcontroller 120 still retains the ability to send alarm data packets to ensure that the management platform can continuously know the device status.

[0046] At a desert photovoltaic power station construction site, a worker was wearing the aforementioned wristband while working. During a lunch break one day, the worker attempted to pry open the wristband's lock using a general-purpose tool. The optical proximity sensor 151 detected a voltage level change the instant the lock was pried open (the tongue displacement exceeded 1 mm). The microcontroller 120 triggered an anti-tamper alarm within milliseconds, sending a "lock open" alarm to the cloud via the LoRa communication module 130, while simultaneously locking the wristband's attendance and location functions. The worker then removed the wristband and had someone else enter the afternoon's work area on his behalf. Because the wristband was locked, the handheld wristband could not generate any valid attendance records, and its location data was no longer uploaded. The cloud management platform 300 displayed the wristband as being in an "abnormally locked" state in real time and recorded the time, location, and type of the alarm. Based on the alarm information, the manager could trace the worker's violation, effectively preventing the disassembly of equipment to evade supervision.

[0047] Based on the above embodiments, as a preferred implementation method, such as Figure 4 As shown, the smart bracelet terminal 100 has a built-in time-sharing positioning state machine, which is configured to include at least the following states: In the high-speed positioning state (S_H), the satellite positioning module 140 operates continuously for a first positioning cycle. Continuous operation means that the RF front-end, baseband processor, and position calculation unit of the satellite positioning module 140 are always powered on, continuously receiving satellite signals from the Global Positioning System (GPS) and the BeiDou Navigation Satellite System, and outputting position coordinates at a frequency of once per minute. In this state, the positioning accuracy can reach a circular error probability (CEP) of 2.5 meters, meaning that 50% of the positioning points fall within a circle with a radius of 2.5 meters centered on the actual location. This state is suitable for situations where workers enter dangerous areas designated by electronic fences, press the SOS button on the wristband, or other emergencies requiring precise and continuous trajectory recording. To control power consumption, the time-division positioning state machine limits the duration of this state, for example, setting a maximum duration of 30 minutes, after which it automatically switches to a low-power positioning state.

[0048] In the low-power positioning state (S_L), the satellite positioning module 140 periodically wakes up and performs positioning at a second positioning period longer than the first positioning period; specifically, the second positioning period is set to 10 minutes. In this state, the satellite positioning module 140 is in sleep mode for most of the time (RF and baseband circuits are powered off), maintaining only a low-power timer. Whenever the timer reaches 10 minutes, it generates a wake-up interrupt. The microcontroller 120 responds to the interrupt and powers on the satellite positioning module 140, initiating a warm-start or cold-start positioning process. The satellite positioning module 140 receives satellite signals and performs position calculations, a process that lasts approximately 10 to 30 seconds. After obtaining valid position coordinates, the microcontroller 120 immediately stores the coordinates in a buffer and powers the satellite positioning module 140 back down to sleep mode. In this state, the average operating current of the satellite positioning module 140 can be controlled below 1.2 mA. Combined with the built-in 3000mAh lithium iron phosphate battery 160, the theoretical battery life is approximately 104 days; considering the power consumption of the pulse wave sensor 110, LoRa communication module 130, and other peripheral circuits, the actual battery life can still exceed 30 days. This state is the default operating mode of the time-sharing positioning state machine, suitable for scenarios where workers are normally constructing within a safe working area.

[0049] In deep sleep mode (S_D), the satellite positioning module 140 is completely shut down; that is, the power supply circuit of the satellite positioning module 140 is cut off by the microcontroller 120 through the power management unit, and the module does not consume any current. Simultaneously, the wristband retains only the low-power monitoring function of the pulse wave sensor 110 (for liveness detection), the receiving channel of the LoRa communication module 130 (for receiving commands from the base station), and the real-time clock circuit. In this state, the total operating current of the wristband can be reduced to below 0.3 mA. This state is suitable for scenarios where workers leave the work area to rest in the living area, or for nighttime shutdowns. When a worker enters the preset electronic fence range of the living area, the system automatically triggers the transition from low-power positioning mode to deep sleep mode; when the worker leaves the living area the next day, the system automatically returns to low-power positioning mode.

[0050] The time-division positioning state machine switches between high-speed positioning, low-power positioning, and deep sleep states according to the current working scenario of the smart bracelet terminal 100. Specifically, the switching rules are as follows: Switching from low-power positioning mode to high-speed positioning mode: The microcontroller 120 performs the switch when it detects any of the following events: (a) the location coordinates output by the satellite positioning module 140 fall within the restricted area defined in the electronic fence; (b) the worker presses the SOS physical button on the wristband; (c) the cloud management platform 300 issues a command to force entry into high-speed positioning mode through the communication base station 200. After the switch, the system immediately and continuously collects location coordinates at a positioning cycle of 1 minute.

[0051] Switching from high-speed positioning to low-power positioning: This switch is performed when any of the following conditions are met: (a) continuous operation in high-speed positioning for more than 30 minutes; (b) the worker's location has left the restricted area and has not entered any restricted area boundary for 5 consecutive minutes; (c) the cloud platform issues a command to restore normal mode. After the switch, the satellite positioning module 140 enters periodic wake-up mode.

[0052] Switching from low-power positioning mode to deep sleep mode: The microcontroller 120 performs the switch when all of the following conditions are met: (a) the location coordinates output by the satellite positioning module 140 fall within the electronic fence of the living area; (b) the real-time clock indicates that the current time is within a preset rest period (e.g., 22:00 to 06:00 the next day); (c) the accelerometer detects that the wristband has been stationary for more than 15 minutes (indicating that the worker has been lying down to rest). After the switch, the satellite positioning module 140 is completely powered off.

[0053] Switching from deep sleep to low-power positioning: This switch is performed when any of the following events are detected: (a) the location coordinates of the satellite positioning module 140 (obtained via timed wake-up, which can be set to briefly wake up every 30 minutes to detect location in deep sleep mode) leave the electronic fence of the living area; (b) the accelerometer detects continuous movement for more than 1 minute; (c) the real-time clock enters a preset start-up period (e.g., 06:30). After the switch, the satellite positioning module 140 resumes 10-minute wake-up positioning.

[0054] To further optimize power consumption management, the time-sharing positioning state machine also includes an adaptive learning algorithm. This algorithm dynamically adjusts the state transition rules based on historical data of workers' daily behavior patterns. Specifically, the system records the high-frequency activity areas and rest periods of workers each day. For example, if the system learns from long-term recording that a worker rests in the living area from 12:00 to 13:00 every day, it will automatically trigger a transition from a low-power positioning state to a deep sleep state at 11:55, and automatically resume the low-power positioning state at 13:05. Through this predictive switching, power consumption is further reduced while ensuring safety. This algorithm does not change the basic logic of the above state transition, but only adds time-dimensional auxiliary judgment to the preset transition conditions.

[0055] For example, during the construction of a desert photovoltaic power station, a worker wearing a wristband was working in the component installation area (safe zone). The wristband was in low-power positioning mode, reporting its location every 10 minutes. At 15:30, the worker needed to enter the area with live equipment (restricted zone) for operation. The wristband's positioning module detected that the position coordinates fell within the polygonal boundary of the restricted zone. The time-sharing positioning state machine immediately switched from low-power positioning mode to high-speed positioning mode, shortening the positioning cycle to 1 minute, continuously recording the worker's movement trajectory after entering the restricted zone. The worker's activities within the restricted zone were monitored in real time with high precision. At 15:50, the worker left the restricted zone. The system detected that the position was now outside the restricted zone, but the high-speed positioning mode would continue for 30 minutes (until 16:00) before automatically switching back to low-power positioning mode. At 19:00 that evening, the worker returned to the living area. The wristband detected that the position fell within the electronic fence of the living area. The time-sharing positioning state machine switched to deep sleep mode, the satellite positioning module 140 was completely shut down, and the wristband entered power-saving mode. At 06:30 the following day, the wristband automatically returned to a low-power positioning state using an adaptive learning algorithm (historical data showed that the worker typically left the living area at 06:30), and the worker wore the wristband to begin the new workday. The entire process required no manual operation from the worker, achieving an automatic balance between positioning accuracy and battery life.

[0056] Based on the above settings, the cumulative operating current of the wristband in low-power positioning mode is approximately 1.2 mA, and the cumulative operating current in deep sleep mode is less than 0.3 mA. Combined with a 3000 mAh lithium iron phosphate battery, the theoretical battery life is up to 104 days. Under comprehensive conditions including the actual operation of communication and sensing modules, the battery life can still exceed 30 days, fully meeting the duration requirements for continuous field operations in desert areas.

[0057] Based on the above embodiments, as a preferred implementation, the time-division positioning state machine further includes an abnormal alarm state (S_A).

[0058] When an anti-tamper alarm, distress signal, or electronic fence boundary crossing event is triggered, the time-division positioning state machine switches to the abnormal alarm state; in the abnormal alarm state, the satellite positioning module 140 operates continuously for a third positioning cycle that is less than the first positioning cycle, and the LoRa communication module 130 sends alarm information to the communication base station 200 at the maximum transmission power.

[0059] The tamper alarm is triggered by the microcontroller 120 when the aforementioned multiple tamper mechanisms detect that the continuous metal wire loop 152 is broken, the pressure distribution pattern deviates from the preset threshold range, or the optical proximity sensor 151 indicates that the latch is open. The distress signal is triggered when the worker presses the SOS physical button on the smart bracelet terminal 100, or actively sends a distress signal through other input methods (such as long-pressing the touch area). The electronic fence boundary crossing event is triggered when the microcontroller 120 determines, based on the real-time location coordinates output by the satellite positioning module 140, that the worker has entered the restricted area (red area) defined by the electronic fence.

[0060] When any of the above events occur, regardless of whether the time-sharing positioning state machine is currently in high-speed positioning, low-power positioning, or deep sleep state, the system will forcibly interrupt the current state and immediately enter the abnormal alarm state.

[0061] The third positioning cycle is shorter than the aforementioned first positioning cycle (i.e., the 1-minute cycle under normal high-speed positioning conditions). Specifically, the third positioning cycle is set to 10 seconds. That is, the satellite positioning module 140 continuously outputs real-time position coordinates at a frequency of once every 10 seconds, which is much higher than the positioning frequency during normal operation, so that the cloud management platform 300 can track the worker's precise location and movement trajectory in real time. In this state, the satellite positioning module 140 maintains continuous operation and does not periodically enter a sleep state.

[0062] The transmission power of the LoRa communication module 130 is increased to its maximum power. According to the system design, the maximum transmission power of the LoRa communication module 130 in the smart bracelet terminal is 20dBm (equivalent to 100 milliwatts). Simultaneously, the microcontroller 120 packages the alarm information and sends it to the communication base station 200 via the LoRa communication module 130 at its maximum transmission power. The alarm information includes at least the following fields: alarm type (tamper alarm / DDoS signal / fence crossing), alarm trigger time, current satellite positioning coordinates (latest output from the satellite positioning module 140), and the unique serial number of the smart bracelet terminal 100.

[0063] In addition, to further improve the reliability of alarm information transmission, the wristband terminal simultaneously sends the alarm information to the communication base station 200 and other nearby base stations, forming a multi-path reporting system to ensure that the alarm information can still be received by the cloud management platform 300 even when some communication links are unstable.

[0064] The abnormal alarm state continues until manual confirmation and processing are completed. That is, after the administrator receives the alarm information through the cloud management platform 300 and completes on-site verification and processing, the platform issues a "clear alarm" command to the corresponding smart bracelet terminal 100. The communication base station 200 forwards the command to the bracelet terminal via the LoRa downlink. After receiving the command, the microcontroller 120 switches the time-division positioning state machine from the abnormal alarm state back to the normal working state. The working state after recovery is determined according to the current scenario: if the bracelet is located in a safe area and there are no other abnormalities, it enters a low-power positioning state; if it is still in a restricted area or requires continuous tracking, it enters a high-speed positioning state.

[0065] Even if the abnormal event itself has ended (e.g., a worker voluntarily retreats from a restricted area to a safe area), the time-sharing positioning state machine remains in an abnormal alarm state until the management personnel manually deactivate it. This design ensures that all abnormal events receive sufficient human attention and closed-loop processing.

[0066] Based on the above embodiments, as a preferred implementation, the cloud management platform 300 or the smart bracelet terminal 100 stores three-level electronic fence configuration information. The three-level electronic fence includes a safe zone, a buffer zone, and a restricted zone, which correspond to work areas with different risk levels.

[0067] Specifically, the safety zone (green zone) is the regular construction work area where workers can move freely. The safety zone is defined using virtual boundaries, described by a polygonal area formed by one or more GPS coordinate points drawn on an electronic map by the cloud management platform administrator. The safety zone typically occupies more than 80% of the entire work area. Within the safety zone, the system operates in a default low-power positioning state, allowing workers to briefly remove their wristbands (e.g., for short breaks) without additional approval.

[0068] The buffer zone (yellow area) is the transition zone between the safe zone and the restricted zone. According to risk management requirements, the buffer zone is set within 2 to 5 meters of the edge of the hazardous area. For example, for an area with live electrical equipment (restricted zone), the ring-shaped area within 2 to 5 meters of its perimeter is defined as the buffer zone. The boundary of the buffer zone is also described by a polygonal ring formed by GPS coordinate points. The purpose of the buffer zone is to provide workers with warning space and prevent them from accidentally entering the restricted zone due to positioning errors or operational negligence.

[0069] Restricted areas (red zones) are areas where unauthorized personnel are explicitly prohibited from entering, including but not limited to areas with electrical equipment, high-altitude work areas, chemical storage areas, and deep foundation pit areas. Restricted areas are established using the principle of least privilege, allowing only authorized personnel of specific trades (such as electrical engineers and certified high-altitude workers) to enter. The boundaries of restricted areas are defined by precise GPS coordinates, and the accuracy of this description directly affects the accuracy of the electronic fence's detection.

[0070] The microcontroller 120 of the smart bracelet terminal 100 is configured as follows: The real-time position coordinates from the satellite positioning module 140 are obtained and denoted as point P(x,y), where x is the longitude value and y is the latitude value. The satellite positioning module 140 outputs coordinates every 10 minutes in low-power positioning mode, every minute in high-speed positioning mode, and every 10 seconds in abnormal alarm mode.

[0071] The type of area where the worker is located is determined based on the real-time location coordinates and the configuration information of the three-level electronic fence.

[0072] Specifically, the electronic fence's judgment logic employs a ray-based polygon inclusion test algorithm. For any point P(x,y) to be detected, a ray is emitted horizontally to the right from that point, and the number of intersections between the ray and the polygon boundary is counted. If the number of intersections is odd, the point is determined to be inside the polygon; if the number is even, the point is determined to be outside the polygon. To eliminate boundary errors, the system uses a 0.5-meter tolerance.

[0073] The microcontroller 120 performs the judgment in the following order: first, it determines whether point P is inside the restricted area polygon; if not, it determines whether it is inside the buffer zone polygon; if still not, it defaults to being in the safe zone.

[0074] Microcontroller 120 performs the judgments in the following priority order: First, determine if point P is located inside any restricted area polygon. If so, the worker's current area type is a restricted area.

[0075] If it is not a restricted area, then determine whether point P is located inside any buffer polygon (or within a buffer ring). If so, the region type is a buffer.

[0076] If it is neither a restricted area nor a buffer zone, the default area type is a safe zone.

[0077] When a worker is detected entering the buffer zone, an early warning is generated. Specifically, the wristband outputs a vibration alert via a vibration motor and displays the warning information on the screen. Simultaneously, the microcontroller 120 sends an "Entering Buffer Zone" alarm record to the cloud management platform 300 via the LoRa communication module 130, including the worker's identity, entry time, and current location coordinates. Furthermore, the system starts a timer; if the worker remains in the buffer zone for more than a preset threshold (e.g., 30 minutes), the alarm level is automatically escalated, and the situation is handled as a restricted area trespass.

[0078] When a worker is detected entering the restricted area, an emergency alarm is generated and output through the sound, light, and vibration module 170 of the smart bracelet terminal 100. Specifically, the bracelet immediately triggers a three-level alarm (sound, light, and vibration are output simultaneously) to strongly remind the worker to evacuate immediately. At the same time, the microcontroller 120 sends the emergency alarm information to the communication base station 200 with the highest priority through the LoRa communication module 130. The base station forwards the alarm to the cloud management platform 300 in real time. After receiving the alarm, the cloud platform pushes an emergency notification to the worker's team leader, safety administrator, and other relevant personnel, and highlights the worker's location and alarm information on the electronic map.

[0079] During the construction of a desert photovoltaic power station, workers wear wristbands to install components within the photovoltaic array area (safe zone). A 3-meter buffer zone is established between this zone and the edge of the safe zone, designated for live equipment.

[0080] One day, a worker gradually approached an area with live electrical equipment. When he entered the buffer zone, his wristband vibrated and displayed a warning message. Realizing he was approaching a danger zone, the worker voluntarily retreated to the safe zone, preventing further alarms from being triggered.

[0081] Another worker, due to operational error, directly entered the area containing live electrical equipment. The wristband's satellite positioning module 140 detected that the coordinates fell within the restricted area polygon, and the microcontroller 120 immediately triggered a level-three alarm (audio-visual and vibration) and simultaneously sent an emergency alarm to the cloud platform. After seeing the worker's real-time location on the platform, the safety administrator quickly used the on-site broadcast system to call for the worker to evacuate, preventing a safety accident. The entire trajectory of the boundary crossing incident was fully recorded for subsequent safety training and accountability.

[0082] Based on the above embodiments, as a preferred implementation, the smart bracelet terminal 100 further includes an offline cache module 180, which is electrically connected to the microcontroller 120.

[0083] The microcontroller 120 is configured to store the generated attendance records and location data in the offline cache module 180 when a communication interruption with the communication base station 200 is detected.

[0084] The offline cache module 180 adopts a circular overwrite storage strategy, and the storage capacity is configured to cache data within at least 72 hours.

[0085] Specifically, the offline cache module 180 is electrically connected to the microcontroller 120 and is used to temporarily store attendance records and location data when the network is interrupted. The total storage capacity of the offline cache module 180 is 128 megabytes, divided into three independent storage areas: an attendance record area (32MB), a location trajectory area (64MB), and a configuration parameter area (32MB). The attendance record area is used to store workers' attendance records, the location trajectory area is used to store continuous location coordinate sequences in time segments, and the configuration parameter area is used to store system parameters such as electronic fence configuration and worker identity templates.

[0086] like Figure 2 and Figure 3 As shown, when a worker first wears the smart bracelet terminal 100, the pressure sensor array 153 on the inner side of the wristband detects the skin contact pressure. The pressure sensor array 153 consists of 16 miniature pressure sensors evenly arranged on the surface of the wristband that contacts the skin, each sensor having a detection range of 0 to 50 kPa. When the detected pressure value exceeds a preset threshold (e.g., ≥5 kPa), the microcontroller 120 determines that the wear is valid and automatically triggers the binding process without manual intervention. At this time, the bracelet's display screen (if any) shows the prompt message "Identity binding in progress, please remain still," and simultaneously activates the dual-wavelength pulse wave sensor 110.

[0087] The microcontroller 120 drives the pulse wave sensor 110 to simultaneously acquire two photoplethysmography (PPG) signals—560 nm green light and 940 nm infrared light—at a sampling rate of 100 Hz for 30 seconds. During the acquisition process, the triaxial accelerometer built into the wristband simultaneously monitors the wearer's motion. If the triaxial accelerometer detects a motion amplitude exceeding a preset threshold (e.g., a rate of change of acceleration greater than 0.5 times the gravitational acceleration per second), it determines that there is a risk of motion artifacts, and the microcontroller 120 automatically extends the acquisition time to 45 seconds to ensure that a sufficiently high-quality signal is obtained for feature extraction.

[0088] After acquisition, the microcontroller 120 performs bandpass filtering on the two PPG signals, with a passband range of 0.5 to 10 Hz. This filtering operation removes power frequency interference (50 / 60 Hz) and baseline drift noise, resulting in a clean pulse wave signal sequence, which serves as the input for subsequent feature extraction.

[0089] The microcontroller 120 calculates features of the preprocessed PPG signal according to the following 10 dimensions to construct a feature vector: f1 Heart rate characteristics: The average heart rate is calculated using a peak detection algorithm, and the unit is beats per minute.

[0090] f2 SDNN: Standard Deviation of Normal-to-Normal Intervals, reflecting the overall variation in heart rate variability.

[0091] f3 RMSSD: Root Mean Square of Successive Differences, reflecting short-term changes in heart rate variability.

[0092] f4 Rise Time: The time it takes for the pulse wave to rise from its trough to its peak, measured in milliseconds.

[0093] f5 Fall time: The time it takes for the pulse wave to go from the peak to the trough, in milliseconds.

[0094] f6 Main peak slope: The slope of the rising edge of the main peak of the pulse wave, reflecting the change in arterial stiffness.

[0095] f7 Dip trough: The amplitude of the second trough of the pulse wave, in millivolts.

[0096] f8 Power Spectrum Entropy: The entropy value of the PPG signal power spectrum, reflecting the complexity of heart rate variability.

[0097] f9 LF: Low-frequency power (0.04–0.15Hz), reflecting sympathetic nerve activity.

[0098] f10 HF: High-frequency power (0.15–0.4 Hz), reflecting parasympathetic nerve activity.

[0099] After normalizing the above 10 feature values, a feature vector with 1 row and 10 columns is formed.

[0100] The microcontroller 120 performs a preliminary verification between the extracted 10-dimensional feature vector and a pre-stored general template. The general template includes the effective range of physiological parameters, such as a heart rate between 40 and 180 beats per minute and an LF / HF ratio between 0.5 and 3.0. If the verification passes (i.e., all feature values ​​are within the normal physiological range), the feature vector is associated with the wearer's employee ID, name, and facial information to generate a unique personal identification template.

[0101] After the template is generated, the microcontroller 120 uses the AES-128 encryption algorithm to encrypt the template data and store it in the non-volatile memory of the wristband to prevent the template from being tampered with or stolen.

[0102] After the template is successfully generated and encrypted, the microcontroller 120 sends a binding request to the cloud management platform 300 via the LoRa communication module 130. The binding request includes the following fields: wristband device ID, wearer's employee ID, and the hash value of the encrypted identity template (e.g., calculated using the SM3 algorithm).

[0103] After receiving the binding request, the cloud management platform 300 verifies the matching relationship between the bracelet device ID and the wearer's employee ID (for example, checking whether the employee ID belongs to a legitimate worker and whether the bracelet has been assigned to another worker). Once verification is successful, the platform binds the bracelet device ID to the wearer's identity information and marks the device binding status as "bound". Simultaneously, the cloud platform returns a confirmation command of successful binding to the smart bracelet terminal 100.

[0104] After receiving the confirmation command via the LoRa communication module 130, the microcontroller 120 marks the locally stored identity template as "locked," meaning it can only be unlocked and used by matching the wearer's PPG feature. This completes the two-way binding between the device and the wearer.

[0105] The microcontroller 120 is also configured to use an SM3 hash chain structure to protect the integrity of the stored data. The hash chain is constructed by using a random number generated when the smart bracelet terminal 100 is first activated as the initial hash value, concatenating the current hash value with the current data record, and then performing an SM3 operation to obtain the next hash value.

[0106] Specifically, when a communication interruption with the communication base station 200 is detected, the generated attendance records and location data are stored in the offline cache module 180. The criteria for determining a communication interruption include: the LoRa communication module 130 fails to receive an acknowledgment response from the base station after sending data multiple times (e.g., three times consecutively), or the real-time detected signal strength is lower than a preset threshold.

[0107] The offline cache module 180 employs a circular overwrite storage strategy. Specifically, the attendance record area uses a First-In-First-Out (FIFO) strategy: newly generated attendance records are written to the storage space sequentially, and when the storage space is full, the latest record automatically overwrites the oldest record. The location tracking area uses a spatial index structure, stored in time-segmented blocks, and also supports circular overwrite. Based on the storage capacity and the average data volume per record, the storage capacity of the offline cache module 180 is configured to cache at least 72 hours of data. That is, in extreme cases without network coverage, the wristband can continuously store all attendance and location data for 3 days without loss.

[0108] To ensure that offline cached data is not tampered with during storage and to guarantee the authenticity and non-repudiation of data uploads after reconnection, the microcontroller 120 is also configured to use an SM3 hash chain structure to protect the integrity of the stored data.

[0109] The SM3 hash chain is constructed as follows: Suppose that when the smart bracelet terminal 100 is first activated, it generates a 64-bit random number as a seed. Define the first value of the hash chain as: chain[0] = SM3(seed) SM3(·) represents the SM3 cryptographic hash algorithm released by the State Cryptography Administration, with an output length of 256 bits.

[0110] For the nth data record (denoted as record[n], where n is numbered starting from 1), the current hash value chain[n] is constructed according to the following formula: chain[n]=SM3(chain[n-1]||record[n]) The "||" operator represents the binary data concatenation operation. Specifically, it concatenates the hash value of the previous record with the current record in sequence to form a new binary string, then performs an SM3 operation on this binary string to obtain the hash value of the current record.

[0111] Using the recursive construction method described above, all data records form an unbreakable chain through hash values. Modifying any record in the chain (e.g., altering attendance time, location coordinates, or deleting a record) will cause all hash values ​​from that record onwards to change, and the continuity of the chain cannot be restored without being detected.

[0112] When the smart bracelet terminal 100 re-enters the coverage area of ​​the communication base station 200 and communication with the communication base station 200 is restored, the microcontroller 120 automatically triggers the data synchronization process.

[0113] The synchronization process includes the following operations: The microcontroller 120 reads the attendance records, location trajectories and corresponding hash chain values ​​(chain[0] to chain[N]) stored in the offline cache module 180.

[0114] The LoRa communication module 130 sends these data in batches to the communication base station 200, which then forwards them to the cloud management platform 300. Upon receiving the data, the cloud management platform 300 first uses the same SM3 algorithm as the wristband and the received initial seed to recalculate the hash value according to the same hash chain construction rules. The recalculated hash chain is compared item by item with the hash chain uploaded by the wristband.

[0115] If the hash chain is completely consistent, the verification passes and the cloud management platform 300 stores the data in the official database; if any inconsistency is found (such as the hash value of the kth record failing to match), it is determined that the data has been tampered with or that an error occurred during transmission. The cloud management platform 300 marks the data segment as "abnormal" and notifies the administrator to conduct a manual verification.

[0116] Based on the above embodiments, as a preferred implementation, the LoRa communication module 130 and the communication base station 200 communicate using a time division multiple access protocol.

[0117] The communication base station 200 or the smart bracelet terminal 100 is configured with frame structure parameters, which divide time into communication frames of equal length, and each communication frame contains multiple time slots.

[0118] The microcontroller 120 is configured to control the LoRa communication module 130 to send data within the time slot allocated to the smart bracelet terminal 100 according to a preset time slot allocation rule.

[0119] Specifically, in order to avoid signal conflicts when multiple smart bracelet terminals 100 send data to the communication base station 200 at the same time, the LoRa communication module 130 and the communication base station 200 communicate using a time division multiple access protocol.

[0120] The communication base station 200 or the smart bracelet terminal 100 is configured with frame structure parameters. These frame structure parameters divide time into communication frames of equal length, each communication frame containing multiple time slots. Specifically, each communication frame has a fixed length and is divided into 16 time slots.

[0121] When each smart bracelet terminal 100 initializes and joins the network, the communication base station 200 assigns it a fixed time slot number. In subsequent communications, the microcontroller 120 is configured to control the LoRa communication module 130 to transmit data within the time slot allocated to the smart bracelet terminal 100, according to a preset time slot allocation rule. That is, each bracelet is only authorized to transmit data within its dedicated time slot period; during other time slots, it remains in a receiving state or a sleep state.

[0122] The smart bracelet terminal 100's LoRa communication cycle is set to once every 5 minutes by default, meaning it sends a data packet (containing location coordinates, attendance records, or heartbeat information) every 5 minutes. Within each communication cycle, the bracelet sends data in its assigned time slot. When the system is in an abnormal alarm state (such as an anti-tamper alarm, SOS distress call, or electronic fence boundary violation), the communication cycle is shortened to once every 30 seconds, and alarm information is sent with the highest priority.

[0123] In addition to TDMA (Time Division Multiple Access), the LoRa communication module 130 also employs an adaptive spreading factor algorithm. The spreading factor (SF) determines the balance between data transmission rate and communication distance. A lower SF value (e.g., SF7) results in a higher data transmission rate but shorter interference resistance and communication distance; a higher SF value (e.g., SF12) results in a lower data transmission rate but longer communication distance and stronger interference resistance.

[0124] The microcontroller 120 continuously monitors the Received Signal Strength Indicator (RSSI) and Signal-to-Noise Ratio (SNR). When the signal strength is good (RSSI above -90dBm), it automatically selects SF7 to improve the transmission rate and reduce the air transmission time, thereby further reducing power consumption. When the signal strength decreases (RSSI below -110dBm) or the packet error rate increases, it automatically switches to a higher spreading factor (such as SF9 or SF12) to ensure that data packets can be successfully received by the base station.

[0125] Based on the above embodiments, as a preferred implementation, the deep learning model is an improved Inception-Time convolutional neural network; The improved Inception-Time convolutional neural network includes: multiple Inception modules connected in series, and a global average pooling layer and a fully connected layer connected after the last Inception module.

[0126] Each Inception module comprises four parallel branches: a 1×1 convolutional layer using a 1×1 kernel, primarily for dimensionality reduction and cross-channel information fusion; a 3×3 convolutional layer using a 3×3 kernel, for extracting medium-scale local features; a 5×5 convolutional layer using a 5×5 kernel, for extracting larger-scale receptive field features; and a 3×3 max-pooling layer using a 3×3 pooling window to perform max-pooling, extracting the most salient features within local regions and reducing the feature map size. The outputs of these four parallel branches are fused through a channel concatenation layer, each outputting feature maps of different scales. Subsequently, the outputs of the four parallel branches are concatenated along the channel dimension (i.e., the channel concatenation layer). Through channel concatenation, feature information of different scales is fused together to output multi-scale features.

[0127] The fully connected layer outputs a 128-dimensional feature vector, which serves as the pulse wave biofeedback vector. The fully connected layer maps the feature vector output by the global average pooling layer to a 128-dimensional feature space through linear transformation and a nonlinear activation function. That is, the fully connected layer outputs a 128-dimensional feature vector, which serves as the pulse wave biofeedback vector.

[0128] During the training phase, the deep learning model employs a contrastive loss function. A training sample pair (a, b) is defined, where a and b are 128-dimensional feature vectors extracted from two PPG signal samples by the network. The label y is defined as follows: if a and b come from the same worker, then y = 1; if they come from different workers, then y = 0. The interval threshold m is set to 0.5. The contrastive loss function is defined as follows: L=(1-y)×max(m-cos(a,b),0)+y×cos(a,b) Where cos(a,b) is the cosine similarity between feature vectors a and b, calculated as: cos(a,b)=(a·b) / (||a||×||b||). This loss function means that for a sample pair (y=1) of the same worker, we want cos(a,b) to be as small as possible (i.e., high similarity, since a smaller cosine similarity indicates greater similarity).

[0129] During the identity verification phase, the wristband collects PPG signals in real time, extracts features locally, and matches them with pre-stored worker templates. A cosine similarity greater than 0.85 is considered successful verification, less than 0.7 is considered a failure, and values ​​in between are subject to manual review. The identity verification result is linked to attendance tracking; only workers who pass verification will have their attendance records accepted by the system.

[0130] The pulse wave feature recognition algorithm of this invention was validated on the public dataset PhysioNet PPG, with an equal error rate (EER) of only 3.2%, which is better than the traditional time-domain feature-based method (EER 8.7%). In actual tests at a desert construction site, considering harsh conditions such as high temperature, sweating, and motion interference, if the error rate rises to 4.8%, it is still within an acceptable range.

[0131] Based on the above embodiments, as a preferred implementation, the satellite positioning module 140 has dual-mode positioning function of Global Positioning System and BeiDou Navigation Satellite System, and is configured to receive and process satellite signals from the Global Positioning System and / or BeiDou Navigation Satellite System.

[0132] Specifically, the satellite positioning module 140 integrates a radio frequency front-end and a baseband processor capable of processing GPS L1 band (1575.42MHz) and BeiDou B1 band (1561.098MHz) signals. After power-on, the module simultaneously searches for available GPS and BeiDou satellites and calculates the current latitude and longitude coordinates, altitude, speed, and positioning accuracy factor (DOP) based on the received satellite signals.

[0133] As described in the foregoing embodiments, the time-division positioning state machine dynamically controls the operating state of the satellite positioning module 140 according to the working scenario. Regardless of whether the satellite positioning module 140 is in high-speed positioning mode, low-power positioning mode, or deep sleep mode, its dual-mode capability remains available. For example: In high-speed positioning mode (positioning cycle of 1 minute), the satellite positioning module 140 operates continuously in dual-mode hybrid mode, receiving all visible GPS and BeiDou satellite signals to obtain optimal positioning accuracy (better than 2.5 meters circular error probability).

[0134] In low-power positioning mode (positioning cycle of 10 minutes), the satellite positioning module 140 is periodically woken up. After being woken up, it quickly positions itself in dual-mode hybrid mode and then immediately goes into sleep mode to balance power consumption and performance.

[0135] In deep hibernation mode, the satellite positioning module 140 is completely shut down and does not receive any satellite signals.

[0136] 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.

[0137] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A wearable, non-intrusive attendance and real-name location management system for migrant workers, characterized in that, This includes smart bracelet terminals, communication base stations, and cloud management platforms; The smart bracelet terminal includes: a pulse wave sensor for attaching to a worker's skin to collect photoplethysmography (PPG) signals, a satellite positioning module for acquiring real-time location coordinates, a LoRa communication module for communicating with a communication base station, and a microcontroller electrically connected to the pulse wave sensor, the satellite positioning module, and the LoRa communication module, respectively. The communication base station is communicatively connected to both the smart bracelet terminal and the cloud management platform. The microcontroller is configured to: The PPG signal is received, and a deep learning model is run to extract the pulse wave biofeature vector from the PPG signal; The extracted pulse wave biometric vector is matched with a pre-stored worker identity template; If a match is successful, the worker's identity will be associated with the real-time location coordinates and / or current timestamp from the satellite positioning module to generate a valid attendance record. The valid attendance records are sent to the communication base station via the LoRa communication module, and then forwarded to the cloud management platform by the communication base station.

2. The wearable, contactless attendance and real-name location management system for migrant workers according to claim 1, characterized in that, The smart bracelet terminal also includes a wristband and multiple anti-tamper mechanisms disposed on the wristband; The multiple anti-tamper mechanism includes: a continuous metal wire loop embedded inside the wristband, a pressure sensor array arranged on the skin contact surface of the wristband, and an optical proximity sensor disposed at the buckle of the wristband. The microcontroller is electrically connected to the continuous metal wire loop, the pressure sensing array, and the optical proximity sensor. When the microcontroller detects that the continuous metal wire loop is disconnected, the pressure distribution pattern collected by the pressure sensing array deviates from the preset threshold range, or the optical proximity sensor detects that the latch is opened, it triggers an anti-tamper alarm, sends the anti-tamper alarm to the cloud management platform through the LoRa communication module, and locks the smart bracelet terminal.

3. The wearable, contactless attendance and real-name positioning management system for migrant workers according to claim 2, characterized in that, The multiple anti-tamper mechanism also includes a cut-resistant wristband body made of Kevlar fiber braided layer, a metal buckle with a double locking mechanism, and an ultra-high frequency radio frequency identification tag embedded inside the wristband; the metal buckle has an asymmetrical keyhole that matches a special irregular key.

4. The wearable, non-intrusive attendance and real-name location management system for migrant workers according to claim 1, characterized in that, The smart bracelet terminal has a built-in time-sharing positioning state machine, which is configured to include at least the following states: In high-speed positioning mode, the satellite positioning module operates continuously for a first positioning cycle. In a low-power positioning state, the satellite positioning module periodically wakes up and performs positioning at a second positioning period that is longer than the first positioning period. In deep sleep mode, the satellite positioning module is completely shut down; The time-division positioning state machine switches between high-speed positioning state, low-power positioning state, and deep sleep state according to the current working scenario of the smart bracelet terminal.

5. The wearable, non-intrusive attendance and real-name positioning management system for migrant workers according to claim 4, characterized in that, The time-division positioning state machine also includes an abnormal alarm state; When an anti-tamper alarm, distress signal, or electronic fence boundary crossing event is triggered, the time-division positioning state machine switches to the abnormal alarm state; in the abnormal alarm state, the satellite positioning module operates continuously for a third positioning cycle that is less than the first positioning cycle, and the LoRa communication module sends alarm information to the communication base station at maximum transmission power.

6. The wearable, contactless attendance and real-name positioning management system for migrant workers according to claim 1, characterized in that, The cloud management platform or the smart bracelet terminal stores three-level electronic fence configuration information, which includes a safe zone, a buffer zone, and a restricted zone. The microcontroller of the smart bracelet terminal is configured as follows: Obtain the real-time position coordinates from the satellite positioning module; The type of area where the worker is located is determined based on the real-time location coordinates and the configuration information of the three-level electronic fence. When it is determined that a worker has entered the buffer zone, an early warning message is generated; When it is determined that a worker has entered the restricted area, an emergency alarm is generated, and an audio-visual vibration alarm is output through the audio-visual vibration module of the smart bracelet terminal.

7. The wearable, contactless attendance and real-name positioning management system for migrant workers according to claim 1, characterized in that, The smart bracelet terminal also includes an offline cache module, which is electrically connected to the microcontroller; The microcontroller is configured to store the generated attendance records and location data in the offline cache module when a communication interruption with the communication base station is detected. The offline caching module adopts a circular overwrite storage strategy, and the storage capacity is configured to cache data for at least 72 hours. The microcontroller is also configured to use an SM3 hash chain structure to protect the integrity of the stored data. The hash chain is constructed by using a random number generated when the smart bracelet terminal is first activated as the initial hash value, concatenating the current hash value with the current data record, and then performing an SM3 operation to obtain the next hash value.

8. The wearable, contactless attendance and real-name positioning management system for migrant workers according to claim 1, characterized in that, The LoRa communication module communicates with the communication base station using a time-division multiple access protocol; The communication base station or the smart bracelet terminal is configured with frame structure parameters, which divide time into communication frames of equal length, and each communication frame contains multiple time slots. The microcontroller is configured to control the LoRa communication module to send data within the time slot allocated to the smart bracelet terminal according to a preset time slot allocation rule.

9. The wearable, contactless attendance and real-name positioning management system for migrant workers according to claim 1, characterized in that, The deep learning model is an improved Inception-Time convolutional neural network; The improved Inception-Time convolutional neural network includes: multiple Inception modules connected in series, and a global average pooling layer and a fully connected layer connected after the last Inception module; Each Inception module includes four parallel branches: a 1×1 convolutional layer, a 3×3 convolutional layer, a 5×5 convolutional layer, and a 3×3 max pooling layer. The outputs of the four parallel branches are fused by a channel splicing layer to output multi-scale features. The fully connected layer outputs a 128-dimensional feature vector, which serves as the biofeature vector of the pulse wave.

10. The wearable, contactless attendance and real-name location management system for migrant workers according to claim 1, characterized in that, The satellite positioning module has dual-mode positioning capabilities for both the Global Positioning System (GPS) and the BeiDou Navigation Satellite System (BDS), and is configured to receive and process satellite signals from the GPS and / or BDS.

Images