A wearable wristwatch system for work personnel area management

Abstract

This invention discloses a wearable wristwatch system for area control of workers, specifically in the field of personnel safety management. It includes: a data acquisition module that synchronously collects heart rate, body temperature, acceleration, and skin contact data of workers using the wristwatch itself, writes a timestamp to each data entry, and outputs a chronologically ordered acquisition sequence; and a positioning and storage module that acquires the positioning data of workers using the wristwatch itself, aligns the positioning data with the acquisition sequence using the same timestamp, and writes it to local storage. By aligning positioning trajectory segments with the acquisition sequence using the same timestamp and generating a candidate event set and event records, and then verifying them with a consistency value and reporting them using a check chain, false alarms can be relatively suppressed and avoided under conditions of positioning drift and network interruption, while retaining verifiable trigger evidence.

Description

Technical Field

[0001] This invention relates to the field of personnel safety management technology, and more specifically, to a wearable wristwatch system for managing work areas. Background Technology

[0002] In wearable wristwatch systems for managing work areas, the mainstream practice in the industry is to address the issues of whether personnel have entered dangerous areas and whether they have experienced any physical abnormalities that require timely alerts. Typically, the wristwatch continuously collects data such as heart rate and body temperature, while also acquiring the personnel's location. The collected data is then compared with preset thresholds or electronic fence boundaries. If the conditions are met, an alarm is triggered and the results are uploaded to the management platform. In scenarios such as mine tunnels, steel structure workshops, or large construction sites, people often move back and forth between floor corridors, doorways, and areas with dense equipment. The positioning signal is easily blocked or reflected, and the network may be intermittent. However, the site requires that an alarm must be triggered immediately if a restricted area is entered or an anomaly occurs. Moreover, the reason for the alarm must be explained afterward. Under these constraints, the mainstream practice will consistently produce two types of directly observable phenomena. One type is that when personnel approach the boundary, their location points jump back and forth, and the system repeatedly alerts them to crossing the boundary within a few seconds to tens of seconds, but the personnel on site do not actually cross the boundary. The other type is that when the watch is loose, sweat contamination, or strenuous exercise causes short-term fluctuations in heart rate and body temperature, the system will also repeatedly alert them to abnormalities. Over time, the personnel on site will treat the alarm as noise and ignore it or even turn off the reminder. In severe cases, they will even take off their watches and put them on the tool cart or give them to others to wear in order to evade control. The reason for these phenomena is that the system often gives an alarm conclusion based on only a single positioning result or a single sampling value, lacking triggering evidence that can be continuously recorded and cross-verified on the wristwatch. It is also difficult to retain the complete triggering process for post-event verification when the network is interrupted. Therefore, the technical problem to be solved by this application is how to ensure that the watch system's alarms for boundary crossings and health abnormalities are neither frequently false alarms nor evaded in work sites where positioning is prone to drift and network interruptions are possible, and how to retain verifiable trigger evidence for each alarm. Summary of the Invention

[0003] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a wearable wristwatch system for area control of workers. By aligning positioning data and health collection data with timestamps to form candidate events and event records, and then generating alarm judgments based on consistency verification and combining verification chains to complete reporting and record keeping, the system solves the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a wearable wristwatch system for area control of workers, comprising: The data acquisition module is used to synchronously collect heart rate data, body temperature data, acceleration data and skin contact data of the workers by the wristwatch itself, write a timestamp to each data acquisition, and output the acquisition sequence arranged in chronological order. The positioning and storage module is used to obtain the positioning data of the operator from the watch body, align the positioning data with the collection sequence according to the same timestamp, and write it into the local storage. At the same time, it generates a check value associated with the previous write for each write to prevent subsequent rewriting, and outputs continuous positioning trajectory segments and a check chain corresponding to each positioning trajectory segment. The candidate event module is used to compare the location trajectory segment with the preset control area boundary to generate boundary crossing candidate events and compare the collected sequence with the preset health alarm rules to generate abnormal candidate events, and output a candidate event set containing event type and event time window; The event logging module is used to extract the positioning trajectory segment within the event time window of each candidate event in the candidate event set, calculate the trajectory continuity index, the wearing stability index, and the motion intensity index, and then merge the above indicators with the corresponding original data segment to form an event record, and output the event record corresponding to each candidate event.

[0005] In a preferred embodiment, it further includes: The verification module is used to perform consistency verification on event records to determine whether the event record meets the alarm conditions. If the alarm conditions are not met, the acquisition frequency is increased or the event time window is extended to regenerate the event record until the preset verification conditions are met or the stop condition is reached. The module outputs the verified alarm event record and the corresponding alarm judgment result. The alarm reporting module is used to trigger audible, visual, or vibration alarms on the watch itself based on the alarm determination result, and report the alarm event record and the corresponding verification chain to the management platform for verification of the alarm triggering process, and output verifiable and traceable alarm records.

[0006] In a preferred embodiment, the acquisition module includes: The watch body reads heart rate data, body temperature data, acceleration data, and skin contact data within the same sampling period. These data are then uniformly marked with the same timestamp and written into a collection sequence. The output is a collection sequence arranged chronologically with all data points aligned. Heart rate data includes an integer or floating-point scalar time series, with units including heartbeats per minute, representing the heart rate value at each timestamp. Body temperature data includes a floating-point scalar time series, with units including degrees Celsius, representing the body surface temperature value at each timestamp. Acceleration data includes a three-dimensional floating-point vector time series, with each timestamp corresponding to three-axis component values, with units including meters per second squared. Skin contact data includes a Boolean or floating-point scalar time series, including effective contact markers or contact mass scores, representing the contact state or intensity between the watch and skin at each timestamp. The watch body determines whether the current wearing is effective based on the changes in skin contact data over multiple consecutive sampling periods. If the wearing is invalid, the heart rate data and body temperature data under the corresponding timestamp are written as invalid data, while the acceleration data and skin contact data under that timestamp are retained, and the collection sequence with a valid wearing mark is output. The watch itself performs gravity-free processing on acceleration data within a continuous time window and calculates the root mean square of the resultant acceleration as the motion intensity. When the motion intensity exceeds the mean plus twice the standard deviation of the motion intensity within that time window for several consecutive times, the sampling frequency of heart rate and body temperature data is increased to twice the current frequency. When the motion intensity falls below the mean plus one standard deviation of the motion intensity for several consecutive times, the sampling frequency is restored to the default sampling frequency. Each change in sampling frequency and the corresponding timestamp are written into the acquisition sequence, and the acquisition sequence with sampling frequency markings is output.

[0007] In a preferred embodiment, the location storage module includes: Within a continuous time window, the sampling time of each location data is read and the timestamp with the smallest time difference from the sampling time is retrieved in the acquisition sequence to form a set of time difference values. The median of the set of time difference values ​​is then taken as the time offset and added to the sampling time of all location data within the continuous time window to complete the alignment by the same timestamp. The aligned location data is then output and written to local storage. Based on the aligned positioning data, sort by timestamp and remove bits from two adjacent positioning data sets to obtain the velocity sequence by the difference between adjacent timestamps. At the same time, take the acceleration data within the same time window from the acquisition sequence and accumulate their absolute values ​​by the difference between timestamps to obtain the velocity change sequence. Then, take the larger of the velocity sequence and the velocity change sequence point by point to form the upper velocity bound sequence. Perform amplitude limiting on each adjacent displacement to ensure that the adjacent displacement does not exceed the product of the corresponding upper velocity bound and the difference between adjacent timestamps to eliminate jump points. Output continuous positioning trajectory segments and write them to local storage.

[0008] In a preferred embodiment, the location storage module further includes: For each written positioning trajectory segment, the byte sequence is concatenated in the order of timestamps. This byte sequence is then concatenated with the previously written check value, and a SHA256 one-way digest operation is performed to obtain the current check value, which is then written to local storage. At the same time, the timestamp and coordinates of the positioning trajectory segment are integerized and stored according to the adjacent difference, and the integerized step size is recorded as the upper bound of the error. This outputs a check chain that corresponds one-to-one with the positioning trajectory segment and is related to the previous and subsequent segments.

[0009] In a preferred embodiment, the candidate event module includes: For each time stamp of the positioning trajectory segment, the signed distance to the boundary of the controlled area is calculated, and the median of the signed distances within the sliding time window is taken as the judgment distance. When the judgment distance changes from not less than zero to less than zero between adjacent time stamps, the next time stamp is recorded as the boundary start time stamp, and when the judgment distance changes from less than zero to not less than zero between adjacent time stamps, the next time stamp is recorded as the boundary end time stamp. Thus, boundary candidate events are output and written into the event time window determined by the boundary start time stamp and the boundary end time stamp. For the acquired sequence, the median of heart rate data and body temperature data are calculated as the baseline within the sliding time window, and the median of the absolute deviation is calculated as the fluctuation scale. At the same time, the resting segment is defined as the timestamp when the sum of acceleration data within the sliding time window is not greater than the median. When the deviation of heart rate data or body temperature data between adjacent timestamps in the resting segment changes from not greater than twice the fluctuation scale to greater than twice the fluctuation scale, it is recorded as the abnormal start timetamp. When the deviation changes from greater than twice the fluctuation scale to not greater than twice the fluctuation scale between adjacent timestamps, it is recorded as the abnormal end timetamp. Thus, abnormal candidate events are output and written into the event time window determined by the abnormal start timetamp and the abnormal end timetamp. Merge out-of-bounds candidate events and abnormal candidate events in chronological order, and concatenate events with an interval of less than the length of the sliding time window into a single event time window and write the event type, then output the candidate event set.

[0010] In a preferred embodiment, the event logging module includes: For each candidate event in the candidate event set, extract the event trajectory from the positioning trajectory segment according to its event time window, and extract the event acquisition segment that matches the event time window timestamp from the acquisition sequence, and output the original data segment corresponding to the candidate event; The displacement of adjacent timestamp coordinates in the event trajectory is calculated and the velocity sequence is obtained by removing the bit and the difference between adjacent timestamps. The ratio of the maximum value to the median of the velocity sequence is taken as the trajectory continuity index. At the same time, the absolute difference between adjacent timestamps of skin contact data in the event acquisition segment is calculated and the median of the absolute difference is taken as the wearing stability index. The system calculates the resultant acceleration from the acceleration data in the event acquisition segment and takes the root mean square of the resultant acceleration within the event time window as the motion intensity index. The system also merges the trajectory continuity index, the wearing stability index, the motion intensity index, and the original data segment with the same candidate event identifier and writes them into the event record. The system outputs the event record that corresponds one-to-one with each candidate event.

[0011] In a preferred embodiment, the verification module includes: Read the trajectory continuity index, wearing stability index and motion intensity index from the event log, multiply the trajectory continuity index and motion intensity index to obtain the upper limit value of motion, and take the reciprocal of the wearing stability index to obtain the wearing confidence value. Then, use the ratio of the upper limit value of motion to the wearing confidence value as the consistency value, and compare the consistency value with the fixed consistency threshold to output the alarm judgment result. When the alarm determination result is no alarm and the consistency value is within the warning range of the fixed consistency threshold, the sampling frequency of heart rate data and body temperature data is increased to twice the current sampling frequency. A new acquisition sequence and a new event record are generated with the new sampling frequency, and the consistency value is recalculated to update the alarm determination result.

[0012] In a preferred embodiment, the verification module further includes: When the alarm determination result is no alarm and the consistency value is lower than the warning range, the event time window corresponding to the event record is extended forward and backward by a fixed extension duration. Based on this, the original data segment is re-extracted to generate a new event record. The consistency value is then recalculated to update the alarm determination result. The update stops when the consistency value remains unchanged for two consecutive calculations or when a sampling frequency increase and a time window extension have been performed once. The verified alarm event record is then output.

[0013] In a preferred embodiment, the alarm reporting module includes: When the alarm determination result is an alarm, an alarm command is generated on the watch body and the sound, light or vibration output is driven. At the same time, the event type and event time window are extracted from the alarm event record, and the check value at the end of the check chain corresponding to the event time window is extracted to form a reporting packet and written into the local queue of reporting code. The reporting packet to be reported is output. The reported packets are numbered according to their generation sequence and reported to the management platform in numerical order when the network is available. If no reply is received from the platform, the reports are repeated at fixed retransmission intervals until a reply is received or the maximum number of retransmissions is reached. After receiving the reply, the reply number is written to local storage to form a closed-loop reporting record, thereby outputting an alarm record that can be verified and traced.

[0014] The technical effects and advantages of this invention are as follows: By aligning the positioning trajectory fragments and the acquisition sequence with the same timestamp to generate a candidate event set and event records, and then verifying them with a consistency value and reporting them in conjunction with a check chain, false alarms can be relatively suppressed and avoided under the conditions of positioning drift and network interruption, while retaining verifiable trigger evidence. By aligning the positioning data with time offset and using acceleration to form an upper limit for velocity to limit adjacent displacements, the repeated boundary crossing warnings caused by the jumping of positioning points near the boundary can be relatively alleviated, and the stability of the boundary crossing candidate event time window can be improved. By determining the effectiveness of wearing the device based on skin contact data and writing heart rate and body temperature data as invalid data during periods of ineffective wearing, the probability of generating abnormal candidate events caused by loose wearing or poor contact can be relatively reduced, thus reducing the triggering of non-real abnormalities. By screening the resting segment by the resultant acceleration and determining the abnormal time window by comparing the deviation between the baseline and the fluctuation scale, the interference of short-term fluctuations in heart rate and body temperature caused by strenuous exercise on the judgment of abnormalities can be relatively reduced, and the specificity of abnormal candidate events can be improved. By extracting raw data segments from candidate events according to event time windows and calculating trajectory continuity indicators, wearing stability indicators, and exercise intensity indicators to form event records, the triggering basis can be written in a structured way and reused by the verification module, thereby improving the interpretability of alarm judgment. By generating verification values ​​that correlate the positioning trajectory segments and retransmitting the reported packet number and recording the receipt, the correspondence between alarm event records and the writing order can be maintained when the network is weak or disconnected and then restored, which facilitates the verification of the alarm process afterward. Attached Figure Description

[0015] Figure 1 This is a layered architecture diagram of the system of the present invention.

[0016] Figure 2 This is a schematic diagram of the system modules of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Refer to the instruction manual appendix Figure 1-2 The present invention provides a wearable wristwatch system for area control of workers; The data acquisition module is used to synchronously collect heart rate data, body temperature data, acceleration data and skin contact data of the workers by the wristwatch itself, write a timestamp to each data acquisition, and output the acquisition sequence arranged in chronological order. The acquisition module is used to form a time-ordered acquisition sequence on the watch body, ensuring that heart rate data, body temperature data, acceleration data, and skin contact data correspond under the same time reference, and maintaining the acquisition results for subsequent processing despite changes in wearing status and exercise status. The acquisition module first completes multi-source reading and writes a unified timestamp within the same sampling period, then generates a valid wearing mark based on skin contact data and invalidates the heart rate and body temperature data during invalid wearing periods. Subsequently, it calculates the exercise intensity based on acceleration data and switches the sampling frequency of heart rate and body temperature data accordingly, recording the frequency changes. This implementation process includes the following steps: The real-time clock of the watch generates a timestamp corresponding to the sampling period. Within the sampling period, the current values ​​of heart rate, body temperature, three-axis components of acceleration, and the current state or mass score of skin contact data are read. The reading results and timestamps are combined to form a collection record and written into the collection sequence to form a collection sequence sorted by timestamp. If any data fails to be read within the sampling period, the data is written as invalid data and marked as invalid. At the same time, the remaining data and timestamps are still written to maintain the timestamp continuity of the collection sequence. The system reads skin contact data corresponding to the most recent fixed number of sampling periods from the acquisition sequence and calculates the effective percentage and fluctuation amplitude. The fixed number, lower limit of effective percentage, and upper limit of fluctuation amplitude are given by preset configuration and stored in the watch body. When the effective percentage is not lower than the lower limit of effective percentage and the fluctuation amplitude is not higher than the upper limit of fluctuation amplitude, the wearing effective mark corresponding to the current timestamp is written as valid, and the heart rate data and body temperature data corresponding to the current timestamp are written into the acquisition sequence according to normal values. When the effective percentage is lower than the lower limit of effective percentage or the fluctuation amplitude is higher than the upper limit of fluctuation amplitude, the wearing effective mark corresponding to the current timestamp is written as invalid, and the heart rate data and body temperature data corresponding to the current timestamp are written as invalid data. At the same time, the acceleration data and skin contact data corresponding to the timestamp are retained. Within a continuous time window, acceleration data undergoes degravity processing, and the resultant acceleration is calculated based on the degravity-free triaxial components. The root mean square of the resultant acceleration is calculated within the continuous time window as the motion intensity, and the mean and standard deviation of the motion intensity are also calculated. The length of the continuous time window, the default sampling frequency, and the sampling frequency multiplication relationship are given by preset configuration and stored in the watch body. When the motion intensity changes from no more than the mean plus twice the standard deviation to more than the mean plus twice the standard deviation between adjacent timestamps, the sampling frequency of heart rate and body temperature data is switched to twice the current sampling frequency. When the motion intensity changes from more than the mean plus one standard deviation to no more than the mean plus one standard deviation between adjacent timestamps, the sampling frequency is restored to the default sampling frequency, and the sampling frequency change and the corresponding timestamp are written into the acquisition sequence to form a sampling frequency marker when a sampling frequency switch occurs. When acceleration data is missing or saturated within the continuous time window, the default sampling frequency is maintained and an invalid motion intensity marker is written. Through the above processing, the acquisition module outputs an acquisition sequence containing timestamps, heart rate data, body temperature data, acceleration data, skin contact data, valid wear markers, and sampling frequency markers. This allows subsequent processing to eliminate the influence of abnormal wear on heart rate and body temperature data based on valid wear markers, and to perform consistent time alignment and windowing of data at different sampling densities based on sampling frequency markers. In practical applications: when the intensity of movement corresponding to acceleration data increases due to the worker's handling or climbing, the sampling frequency of heart rate and body temperature data is increased, and the sampling frequency marker is recorded to retain the sampling results during the change in movement state. When the watch body becomes loose, causing the effective proportion of skin contact data to decrease or the fluctuation range to increase within a fixed number of sampling periods, the valid wear marker is written as invalid, and the heart rate and body temperature data with the corresponding timestamp are written as invalid data to avoid numerical fluctuations caused by abnormal wear in subsequent judgments.

[0019] The positioning and storage module is used to obtain the positioning data of the operator from the watch body, align the positioning data with the collection sequence according to the same timestamp, and write it into the local storage. At the same time, it generates a check value associated with the previous write for each write to prevent subsequent rewriting, and outputs continuous positioning trajectory segments and a check chain corresponding to each positioning trajectory segment. The positioning storage module is used to establish a unified time base for positioning data and acquisition sequences on the watch body and write them to local storage. This allows subsequent processing to read positioning data and acquisition sequences at the same timestamp, while maintaining the continuity of positioning trajectory segments and the verifiability of stored content in scenarios such as positioning jumps and post-rewrites. The positioning storage module first estimates the time offset within a continuous time window and aligns the timestamps of the positioning data and acquisition sequences accordingly. Then, it uses acceleration data to form a velocity upper bound constraint to limit and correct the aligned positioning data and output continuous positioning trajectory segments. Subsequently, it generates a check value associated with the previous write for each written positioning trajectory segment and performs differential integerization storage to control storage overhead. This implementation process includes the following steps: Within a continuous time window, the sampling time of each location data point is read, and a set of timestamps is read from the acquisition sequence. For each location data point, the time difference between its sampling time and each timestamp is calculated, and the time difference with the smallest absolute value is selected as the time difference value of that location data point. The time difference values ​​of all location data points within the continuous time window are combined into a time difference value set, and the median of the time difference value set is taken as the time offset. The time offset is added to the sampling time of all location data points within the continuous time window to obtain aligned location data, which is then written to local storage in timestamp order. When the number of location data points within a continuous time window is insufficient to form a time difference value set, the time offset obtained from the previous continuous time window remains unchanged, and the aligned location data is written to local storage. At the same time, the time offset is marked for subsequent reading. The aligned positioning data is sorted by timestamp, and the coordinate difference between two adjacent positioning data is calculated to obtain adjacent displacements. Then, the velocity sequence is obtained by removing adjacent bits and using the adjacent timestamp difference. At the same time, the acceleration data corresponding to the timestamp within the same time window is read from the acquisition sequence, and the absolute value of the acceleration data is taken and accumulated according to the timestamp difference to obtain the velocity change sequence. Then, the larger value is taken point by point between the velocity sequence and the velocity change sequence to form the upper limit sequence of velocity. The amplitude is limited for each adjacent displacement so that the adjacent displacement does not exceed the product of the corresponding upper limit of velocity and the adjacent timestamp difference to obtain the corrected displacement. Based on the corrected displacement, the coordinates of each timestamp are reconstructed to form a continuous positioning trajectory segment and written to local storage. When there are invalid markers in the acceleration data within the same time window, the velocity change accumulation is skipped by the timestamp difference corresponding to the invalid marker, and only the velocity sequence is used to generate the upper limit sequence of velocity. At the same time, an invalid acceleration marker is written for subsequent reading. For each written positioning trajectory segment, the timestamps and corresponding coordinates are concatenated in time to form a byte sequence, which is then concatenated with the previously written checksum. A SHA-256 one-way digest operation is then performed to obtain the current checksum, which is written to local storage. Simultaneously, the timestamps and coordinates of the positioning trajectory segment are integerized based on adjacent differences and quantized with a preset integerization step size before being written to local storage. The integerization step size is used as an upper bound for error and written to local storage to characterize the quantization error. This outputs a checksum chain that corresponds one-to-one with the positioning trajectory segments and is interconnected. If the previously written checksum does not exist, a sequence of all-zero bytes is used as the previously written checksum and written to the checksum chain start marker. Through the above processing, the positioning storage module completes the timestamp alignment of positioning data and acquisition sequence within a continuous time window and outputs continuous positioning trajectory segments. This allows the candidate event module to read the positioning trajectory segments and acquisition sequence with a unified timestamp. It also suppresses the impact of positioning jump points on boundary judgments through velocity upper bound constraints. Simultaneously, it associates adjacent writes through a verification chain to form correlated verification values ​​and reduces local storage usage through differential integerization. In practical applications: when workers enter an indoor obstructed area causing a fixed offset between the positioning data sampling time and the acquisition sequence timestamp, the median of the time difference set within the continuous time window is used to obtain the time offset and write it to local storage after aligning the positioning data. When indoor multipath causes a single large displacement jump in the aligned positioning data, the product of the velocity upper bound sequence and the difference between adjacent timestamps is used to limit adjacent displacements and form continuous positioning trajectory segments. When the management platform verifies alarm event records, the correlation between the current verification value corresponding to the positioning trajectory segment and the previously written verification value is used to verify whether the writing order of the positioning trajectory segment has been rewritten.

[0020] The candidate event module is used to compare the location trajectory segment with the preset control area boundary to generate boundary crossing candidate events and compare the collected sequence with the preset health alarm rules to generate abnormal candidate events, and output a candidate event set containing event type and event time window; The candidate event module is used to extract time intervals that may trigger alarms from positioning trajectory segments and acquisition sequences on the watch body and form a candidate event set. This allows the subsequent event recording module to extract raw data segments according to event time windows and calculate trajectory continuity indicators, wearing stability indicators, and exercise intensity indicators. The candidate event module first generates boundary crossing candidate events based on the spatial relationship between the positioning trajectory segments and the boundary of the controlled area and determines the boundary crossing start time stamp and boundary crossing end time stamp. Then, it generates abnormal candidate events based on the heart rate data and body temperature data of the acquisition sequence and determines the abnormality start time stamp and abnormality end time stamp. Subsequently, the boundary crossing candidate events and abnormal candidate events are merged in chronological order, and when the time interval meets the splicing condition, a single event time window is formed and written into the event type. This implementation process includes the following steps: To determine the event time window for boundary crossing candidate events, the signed distance from each timestamp coordinate in the positioning trajectory segment to the boundary of the controlled area is calculated. The signed distance is negative when the coordinate is outside the controlled area and non-negative when it is inside the controlled area. The median of the signed distances within the sliding time window is used to obtain the judgment distance. When the judgment distance changes from not less than zero to less than zero between adjacent timestamps, the next timestamp is written as the boundary crossing start timestamp and the event type is written as the boundary crossing candidate event. When the judgment distance changes from less than zero to not less than zero between adjacent timestamps, the next timestamp is written as the boundary crossing end timestamp. The event time window is determined by the boundary crossing start timestamp and the boundary crossing end timestamp and then written as the boundary crossing candidate event. When the positioning trajectory segment has missing coordinates or invalid markers within the sliding time window, the median of the signed distances corresponding to the valid coordinates within the sliding time window is calculated. If the number of valid coordinates is less than the minimum number given by the preset configuration, the boundary crossing judgment of that sliding time window is skipped. To determine the event time window for abnormal candidate events, heart rate data and body temperature data are extracted from the acquired sequence within the sliding time window, and the median of the heart rate data and body temperature data are calculated as the baseline. The median of the absolute deviation is also calculated as the fluctuation scale. At the same time, the combined acceleration is calculated from the acceleration data within the same sliding time window, and the timestamps where the combined acceleration is not greater than the median are taken as the resting segment. During the resting period, the deviation of heart rate data from the heart rate baseline and the deviation of body temperature data from the body temperature baseline are calculated according to the timestamp and compared with twice the corresponding fluctuation scale. When any deviation changes from no more than twice the corresponding fluctuation scale to more than twice the corresponding fluctuation scale between adjacent timestamps, the next timestamp is written as the abnormal start timestamp and the event type is written as an abnormal candidate event. When any deviation changes from more than twice the corresponding fluctuation scale to no more than twice the corresponding fluctuation scale between adjacent timestamps, the next timestamp is written as the abnormal end timestamp. After determining the event time window with the abnormal start timestamp and the abnormal end timestamp, the abnormal candidate event is written. When there are invalid markers in the heart rate data or body temperature data within the sliding time window, skip the timestamps corresponding to the invalid markers and do not generate abnormal candidate events when the number of valid data is less than the minimum number given by the preset configuration. To form a candidate event set, out-of-bounds candidate events and abnormal candidate events are sorted by their start timestamps and written into the candidate event set in sequence. When the time interval between the start timestamp of the latter and the end timestamp of the former of two adjacent candidate events is less than the length of the sliding time window, the end timestamp of the former candidate event is updated to the end timestamp of the latter candidate event and the event type is written as a composite event type to form a single event time window. In other cases, the event time windows of each candidate event remain unchanged and the corresponding event types are written, thereby outputting a candidate event set containing event types and event time windows. Through the above processing, the candidate event module unifies the boundary changes of the positioning trajectory segment and the abnormal changes of the acquisition sequence into a candidate event set with event type and event time window, so that the event recording module can extract the event trajectory and event acquisition segment one by one according to the candidate event set and form an event record. At the same time, the interference of positioning jump point and motion state on the generation of candidate events is reduced by median smoothing of sliding time window and rest segment filtering. In practical applications: When workers walk along the boundary of the controlled area, causing the signed distance of the positioning trajectory segment to fluctuate near zero, the judgment distance within the sliding time window is used to suppress single-point fluctuations and the time of sign change is written into the boundary start time stamp and boundary end time stamp. When workers run for a short time, causing the heart rate data to rise but the sum of acceleration data to rise at the same time, the resting segment screening prevents this time interval from entering the abnormal candidate event generation process. When the time interval between the boundary candidate event end time stamp and the abnormal candidate event start time stamp is less than the length of the sliding time window, the candidate event set will splice the two into a single event time window and write it into the composite event type so that it can be verified according to the same time window later.

[0021] The event logging module is used to extract the positioning trajectory segment within the event time window of each candidate event in the candidate event set, calculate the trajectory continuity index, the wearing stability index, and the motion intensity index, and then merge the above indicators with the corresponding original data segment to form an event record, and output the event record corresponding to each candidate event.

[0022] The event logging module converts the candidate event set into event records that can be directly used for consistency verification. This allows the verification module to read the event time window, raw data segments, and trajectory continuity, wearing stability, and exercise intensity indices calculated from the raw data segments under the same candidate event identifier. The event logging module first extracts the corresponding event trajectory and event acquisition segments from the positioning trajectory segments and acquisition sequences according to the event time window to form raw data segments. Then, it calculates the velocity sequence based on the event trajectory to obtain the trajectory continuity index. Simultaneously, it obtains the wearing stability index based on skin contact data. Subsequently, it obtains the exercise intensity index based on acceleration data and merges each index with the raw data segments into an event record. This implementation process includes the following steps: To establish the correspondence between candidate events and raw data segments, for each candidate event in the candidate event set, its event type and event time window are read, and the start and end timestamps of the event time window are used as indices. The timestamp coordinate sequence within the event time window is extracted from the positioning trajectory segment to form the event trajectory. At the same time, heart rate data, body temperature data, acceleration data, and skin contact data with the same timestamps are extracted from the acquisition sequence to form the event acquisition segment. The event trajectory and the event acquisition segment are written into the raw data segment according to the same event time window and a candidate event identifier is written for subsequent reading. When there is a missing positioning trajectory or a missing acquisition sequence within the event time window, a missing timestamp is written with a missing marker, and the existing timestamp data is retained to keep the boundary of the event time window unchanged. To extract trajectory continuity and wearing stability indices from raw data fragments, the event trajectory is sorted by timestamp, and adjacent displacements are calculated for adjacent timestamp coordinates. Then, the velocity sequence is obtained by removing adjacent positions and calculating the difference between adjacent timestamps. The ratio of the maximum value to the median of the velocity sequence is taken as the trajectory continuity index and written into the index field corresponding to the candidate event identifier. Simultaneously, skin contact data is read from the event collection fragments, and the absolute difference between adjacent timestamps is calculated. The median of the absolute differences is taken as the wearing stability index and written into the index field corresponding to the candidate event identifier. When the median of the velocity sequence is zero or nonexistent, the trajectory continuity index is written to the maximum indication value given by the preset configuration, and a velocity sequence unavailable flag is added. When invalid flags exist in the skin contact data, the timestamps corresponding to the invalid flags are skipped, and if the number of valid data is insufficient, the wearing stability index is written as invalid, and a skin contact data unavailable flag is added. To extract motion intensity indicators from raw data segments and form event records, acceleration data is read for each event acquisition segment, and the resultant acceleration at each timestamp is calculated. The root mean square of the resultant acceleration within the event time window is then used as the motion intensity indicator and written into the indicator field corresponding to the candidate event identifier. The trajectory continuity indicator, wearing stability indicator, motion intensity indicator, and raw data segment are merged and written into an event record according to the same candidate event identifier and written to local storage for the verification module to read, thereby outputting an event record corresponding to each candidate event. When there are invalid markers in the acceleration data, the timestamps corresponding to the invalid markers are skipped, and if the number of valid data is insufficient, the motion intensity indicator is written as invalid and an acceleration data unavailable marker is written. Through the above processing, the event logging module maps each candidate event in the candidate event set to an event record containing original data fragments and three types of indicators, enabling the verification module to complete consistency value calculation and alarm determination under the same candidate event identifier, and maintain the correspondence between event time windows and data validity through missing and unavailable markers; In practical applications: When the event time window of a cross-boundary candidate event covers the interval where a single jump point occurs in the positioning trajectory segment, the ratio of the maximum value to the median of the velocity sequence increases and is written as a trajectory continuity index. When the watch body is loosened, causing drastic changes in skin contact data between adjacent timestamps, the median of the absolute difference in skin contact data increases and is written as a wearing stability index. When the operator is running within the event time window, the root mean square of the resultant acceleration increases and is written as a motion intensity index. The above indicators, together with the original data segment, are written into the event record and then read by the verification module to output the alarm judgment result.

[0023] The verification module is used to perform consistency verification on event records to determine whether the event record meets the alarm conditions. If the alarm conditions are not met, the acquisition frequency is increased or the event time window is extended to regenerate the event record until the preset verification conditions are met or the stop condition is reached. The module outputs the verified alarm event record and the corresponding alarm judgment result. The verification module is used to verify the consistency of event records and output alarm judgment results, enabling the alarm reporting module to trigger alarms and report alarm event records based on the alarm judgment results. The verification module first reads the trajectory continuity index, wearing stability index, and exercise intensity index from the event records and calculates the consistency value. Then, it compares the consistency value with a fixed consistency threshold and warning interval to obtain the initial alarm judgment result. When the initial alarm judgment result is no alarm, it selects to increase the sampling frequency of heart rate data and body temperature data or expand the event time window according to the interval where the consistency value is located to generate new event records and update the consistency value, until the stopping condition is met and the alarm event record is output. This implementation process includes the following steps: To obtain the initial alarm judgment result, the trajectory continuity index, wearing stability index, and exercise intensity index are read from the event log. The trajectory continuity index and exercise intensity index are multiplied to obtain the upper limit value of the exercise. At the same time, the reciprocal of the wearing stability index is taken to obtain the wearing confidence value. The upper limit value of the exercise is then divided by the wearing confidence value to obtain the consistency value. The fixed consistency threshold is given by the preset configuration and stored in the watch body. When the consistency value is greater than or equal to the fixed consistency threshold, the alarm judgment result is output as an alarm. When the consistency value is less than the fixed consistency threshold, the alarm judgment result is output as no alarm. The consistency value and the alarm judgment result are written to the verification field corresponding to the event log. When the wearing stability index is zero or written as invalid, the wearing confidence value is written as the minimum positive value given by the preset configuration and a wearing stability index unavailable mark is written to avoid abnormal reciprocal calculation. To supplement sampling information when the consistency value approaches the fixed consistency threshold, the warning interval is determined by the fixed consistency threshold and the preset interval width. The preset interval width is given by a preset configuration and stored in the watch body. The warning interval is the closed interval between the fixed consistency threshold minus the preset interval width and the fixed consistency threshold. When the alarm judgment result is no alarm and the consistency value falls into the warning interval, the sampling frequency of heart rate data and body temperature data is switched from the current sampling frequency to twice the current sampling frequency and written to the sampling frequency mark. A new acquisition sequence is generated according to the new sampling frequency, and a new event record corresponding to the event record is regenerated accordingly. Then, the trajectory continuity index, wearing stability index, and exercise intensity index are read from the new event record, and the consistency value is recalculated to update the alarm judgment result. When the watch body shows continuous invalid heart rate data or body temperature data after switching the sampling frequency, it is restored to the default sampling frequency and a sampling frequency recovery mark is written, while keeping the alarm judgment result of this update as no alarm. To supplement time range information when the consistency value is significantly lower than the warning range, when the alarm judgment result is no alarm and the consistency value is less than the lower limit of the warning range, the event time window corresponding to the event record is extended forward and backward by a fixed extension duration. The fixed extension duration is given by a preset configuration and stored in the watch body. Based on the extended event time window, the positioning trajectory segment and the acquisition sequence are re-extracted to form a new original data segment and a new event record is regenerated. Then, the trajectory continuity index, wearing stability index, and exercise intensity index are read from the new event record and the consistency value is recalculated to update the alarm judgment result. When the consistency value remains unchanged for two consecutive calculations or when one sampling frequency switch and one event time window extension have been performed, the update stops. The final consistency value and the final alarm judgment result are written into the alarm event record and the verified alarm event record is output. When the extended event time window exceeds the coverage of the locally stored positioning trajectory segment or acquisition sequence, the extended event time window is truncated according to the available boundary and a truncation mark is written. Through the above processing, the verification module outputs the alarm judgment result under the fixed consistency threshold, and when the consistency value is close to the fixed consistency threshold, it supplements the time density of heart rate data and body temperature data by increasing the sampling frequency. When the consistency value is significantly low, it supplements the time range of the positioning trajectory segment and the acquisition sequence by expanding the event time window, thereby updating the event record and outputting the alarm event record under the same candidate event identifier. In practical applications: When an out-of-bounds candidate event occurs in the jitter range of the positioning trajectory segment boundary and the wearing stability index is normal, the consistency value may fall into the warning range. The watch body will increase the sampling frequency of heart rate data and body temperature data to twice and generate new event records to update the consistency value. When the event time window of the abnormal candidate event is short and there is a short-term missing acceleration data that causes the exercise intensity index to be unstable, the consistency value may be lower than the lower limit of the warning range. The event time window is extended forward and backward for a fixed duration and new event records are regenerated to update the consistency value. Finally, the update ends and the alarm event record is output for the alarm reporting module to read, triggered by the stopping condition.

[0024] The alarm reporting module is used to trigger audible, visual, or vibration alarms on the watch itself based on the alarm judgment result, and report the alarm event record and the corresponding verification chain to the management platform for verification of the alarm triggering process. It outputs verifiable and traceable alarm records. The alarm reporting module is used to trigger alarms on the watch body based on the alarm judgment result and report the alarm event record and verification chain association information to the management platform. This allows the management platform to locate the corresponding alarm process according to the event type and event time window, and to verify the alarm event record and local writing order based on the verification value at the end of the verification chain. The alarm reporting module first generates an alarm command when the alarm judgment result is an alarm and drives the sound, light or vibration output. At the same time, it extracts the fields required for reporting from the alarm event record to form a reporting packet and writes it to the local queue. Then, it numbers the reporting packets in the local queue according to the generation order and reports them in numerical order when the network is available. It also performs retransmission control and acknowledgment disking based on the platform's feedback. This implementation process includes the following steps: When the alarm determination result is an alarm, an alarm command containing alarm type, alarm duration and alarm intensity parameters is generated on the watch body and the sound, light or vibration output is driven. At the same time, the event type and event time window are read from the alarm event record and the check value at the end of the check chain corresponding to the event time window is read from the check chain. The event type, event time window, check value at the end of the check chain and the alarm command identifier generated this time are concatenated to form a reporting packet and written to the local queue for subsequent reading. The reporting packet to be reported is output. When the check value at the end of the check chain is missing, the check value at the end of the check chain is written as an all-zero byte sequence and a check chain missing mark is written. At the same time, a reporting packet is still generated and written to the local queue. For each report packet written to the local queue, an incrementing number is generated according to the writing order and written to the header of the report packet. When network availability is detected, the report packets are read sequentially from smallest to largest number and reported to the management platform, and a receipt timer is started. The fixed retransmission interval and the maximum number of retransmissions are given by a preset configuration and stored in the watch itself. When the receipt timer expires and no receipt is received from the management platform, the report is repeated at the fixed retransmission interval until a receipt is received or the maximum number of retransmissions is reached. When a receipt is received, the receipt number is read from the receipt and written to the local storage along with the corresponding report packet number to form a closed-loop report record. When the maximum number of retransmissions is reached and no receipt is received, the report packet is marked as unacknowledged and kept in the local queue to wait for the next time the network is available to be re-reported. Through the above processing, when the alarm determination result is an alarm, the alarm reporting module synchronously triggers the alarm output on the wristwatch and generates a reporting packet containing the event type, event time window, and verification value at the end of the verification chain, which is written to the local queue. When the network is available, the packet is reported in numerical order and retransmission control and acknowledgment are performed based on the receipt, thus forming a closed-loop reporting record corresponding to the alarm event record on the management platform. In practical applications: when a worker enters a restricted area and triggers an alarm determination result, the sound, light, or vibration output is immediately activated to prompt the worker to leave the restricted area. At the same time, the reporting packet is written to the local queue and reported to the management platform in numerical order after the network is restored. After the management platform returns the receipt, the receipt number and the reporting packet number are written to the local storage to correspond to the alarm event record. When the on-site network remains unavailable, causing the maximum number of retransmissions to be reached, the reporting packet remains in an unacknowledged state and waits for the network to become available again before continuing to report.

[0025] like Figure 1 As shown, it should be noted that the architecture of this system is organized from top to bottom as follows: edge acquisition layer, positioning and storage layer, candidate event layer, event recording layer, verification and alarm layer, and platform interaction layer. The acquisition module reads heart rate data, body temperature data, acceleration data and skin contact data in the same sampling cycle and writes them into timestamps to form an acquisition sequence that is arranged in chronological order and can be aligned. The positioning storage module performs timestamp alignment on the positioning data and generates continuous positioning trajectory segments after suppressing positioning jump points with the upper limit of speed. At the same time, it uses a check chain to represent the correlation check value between the front and back of the positioning trajectory segments and writes the positioning trajectory segments and check chains into local storage. The local storage also saves the receipt number for subsequent reporting and closed-loop verification. The candidate event module calculates boundary crossing candidate events for the positioning trajectory segment based on the boundary of the control area, and calculates abnormal candidate events for the collected sequence according to the health alarm rules and related configurations. The health alarm rules and related configurations are used to give the calculation caliber of the baseline and fluctuation scale and determine the screening caliber of the resting segment. Based on this, the candidate event module outputs a set of candidate events containing event type and event time window. The event recording module extracts raw data segments from the positioning trajectory segment and the acquisition sequence according to the event time window, and calculates the trajectory continuity index to characterize the degree of change in the event trajectory, calculates the wearing stability index to characterize the degree of fluctuation in the skin contact state, and calculates the motion intensity index to characterize the level of acceleration change. Then, the above indicators are merged with the raw data segments to form the event record. The verification module reads the event log and calculates the consistency value under the constraints of the verification-related configuration. The upper bound value of motion is obtained by multiplying the trajectory continuity index and the motion intensity index. The wearing confidence value is obtained by taking the reciprocal of the wearing stability index. The consistency value is represented by the ratio of the upper bound value of motion to the wearing confidence value. It is compared with the consistency threshold to output the alarm judgment result and form an alarm event record. The verification-related configuration includes the consistency threshold, the warning interval width, the extension duration, the retransmission interval and the number of times to limit the consistency judgment and recalculation strategy. When the alarm judgment result is in the warning interval, the sampling frequency is increased to return to the acquisition module to update the acquisition sequence and recalculate the consistency value. When the alarm judgment result is lower than the warning interval, the time window is extended and recalculation is performed to return to the event log module to update the event log and recalculate the consistency value. When the alarm determination result is an alarm, the alarm reporting module generates an alarm command and drives an audible or visual alarm or vibration alarm. At the same time, it extracts the event type and event time window from the alarm event record and binds the verification value at the end of the verification chain to form a reporting packet. The reporting packets are sent to the management platform in numerical order. The management platform returns a receipt for receipt correspondence and retransmission control. The receipt number is written to local storage to form a traceable reporting closed-loop record.

[0026] Working principle: The watch body simultaneously collects heart rate data, body temperature data, acceleration data and skin contact data at a fixed sampling period and writes them into a timestamp to form a collection sequence. At the same time, it acquires positioning data and aligns it with the collection sequence with the same timestamp before writing it into local storage. When writing positioning trajectory segments, it generates corresponding check values ​​to form a check chain. Subsequently, the location trajectory segment is compared with the boundary of the controlled area to obtain the event time window of the boundary crossing candidate event. The changes in heart rate data and body temperature data in the resting period of the collected sequence are compared to obtain the event time window of the abnormal candidate event, and they are merged to form a candidate event set. For each candidate event, the event trajectory and event collection segment are extracted according to the event time window to form the original data segment. After calculating the trajectory continuity index, wearing stability index, and exercise intensity index, an event record is generated. The verification module calculates the consistency value based on this and compares it with a fixed consistency threshold to obtain the alarm judgment result. If necessary, the event record is regenerated by increasing the sampling frequency or expanding the event time window and then verified again. Finally, the alarm event record is output. When an alarm is determined, the watch triggers an audible and visual alarm or vibration alarm and sends the alarm event record and the verification value at the end of the verification chain to the control platform as a reporting package and records the receipt.

[0027] In practical applications, taking the control of restricted areas in a factory as an example, after workers enter the site wearing wristwatches, the wristwatches continuously record the data collection sequence and location trajectory fragments, generating a verification chain. When a worker crosses a boundary, the start and end timestamps of the boundary crossing are recorded, forming a boundary crossing candidate event. When a worker stops to rest, if the heart rate or body temperature data shows a significant jump relative to the baseline, an abnormal start and end timestamps are recorded, forming an abnormal candidate event. The system extracts raw data from the time window of the candidate events and calculates three types of indicators to generate event records. It then calculates a consistency value to determine whether to issue an alarm. If the information is insufficient, the sampling frequency is increased or the time window is appropriately extended before re-verification. Once an alarm is confirmed, the wristwatch immediately vibrates or flashes to alert the worker. At the same time, it sends a reporting packet to the control platform according to the network status and writes the receipt number to local storage after receiving the receipt to correspond to the alarm record.

[0028] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A wearable wristwatch system for area control of workers, characterized in that, include: The data acquisition module is used to synchronously collect heart rate data, body temperature data, acceleration data and skin contact data of the workers by the wristwatch itself, write a timestamp to each data acquisition, and output the acquisition sequence arranged in chronological order. The positioning and storage module is used to obtain the positioning data of the operator from the watch body, align the positioning data with the collection sequence according to the same timestamp, and write it into the local storage. At the same time, it generates a check value associated with the previous write for each write, and outputs continuous positioning trajectory segments and a check chain corresponding to each positioning trajectory segment. The candidate event module is used to compare the location trajectory segment with the preset control area boundary to generate boundary crossing candidate events and compare the collected sequence with the preset health alarm rules to generate abnormal candidate events, and output a candidate event set containing event type and event time window; The event logging module is used to extract the positioning trajectory segment within the event time window of each candidate event in the candidate event set, calculate the trajectory continuity index, the wearing stability index, and the motion intensity index, and then merge the above indicators with the corresponding original data segment to form an event record, and output the event record corresponding to each candidate event.

2. The wearable wristwatch system for area control of workers according to claim 1, characterized in that: Also includes: The verification module is used to perform consistency verification on event records to determine whether the event record meets the alarm conditions. If the alarm conditions are not met, the acquisition frequency is increased or the event time window is extended to regenerate the event record until the preset verification conditions are met or the stop condition is reached. The module outputs the verified alarm event record and the corresponding alarm judgment result. The alarm reporting module is used to trigger audible, visual, or vibration alarms on the watch itself based on the alarm determination result, and report the alarm event record and the corresponding verification chain to the management platform for verification of the alarm triggering process, and output verifiable and traceable alarm records.

3. A wearable wristwatch system for area control of workers according to claim 2, characterized in that: The acquisition module includes: The watch body reads heart rate data, body temperature data, acceleration data and skin contact data in the same sampling period, and writes each data in the sampling period into the acquisition sequence after being uniformly marked with the same timestamp. The output is an acquisition sequence arranged in chronological order and each data can be aligned one by one. The watch body determines whether the current wearing is effective based on the changes in skin contact data over multiple consecutive sampling periods. If the wearing is invalid, the heart rate data and body temperature data under the corresponding timestamp are written as invalid data, while the acceleration data and skin contact data under that timestamp are retained, and the collection sequence with a valid wearing mark is output. The watch itself performs gravity-free processing on acceleration data within a continuous time window and calculates the root mean square of the resultant acceleration as the motion intensity. When the motion intensity exceeds the mean plus twice the standard deviation of the motion intensity within that time window for several consecutive times, the sampling frequency of heart rate and body temperature data is increased to twice the current frequency. When the motion intensity falls below the mean plus one standard deviation of the motion intensity for several consecutive times, the sampling frequency is restored to the default sampling frequency. Each change in sampling frequency and the corresponding timestamp are written into the acquisition sequence, and the acquisition sequence with sampling frequency markings is output.

4. A wearable wristwatch system for area control of workers according to claim 3, characterized in that: The location storage module includes: Within a continuous time window, the sampling time of each location data is read and the timestamp with the smallest time difference from the sampling time is retrieved in the acquisition sequence to form a set of time difference values. The median of the set of time difference values ​​is then taken as the time offset and added to the sampling time of all location data within the continuous time window to complete the alignment by the same timestamp. The aligned location data is then output and written to local storage. Based on the aligned positioning data, sort by timestamp and remove bits from two adjacent positioning data sets to obtain the velocity sequence by the difference between adjacent timestamps. At the same time, take the acceleration data within the same time window from the acquisition sequence and accumulate their absolute values ​​by the difference between timestamps to obtain the velocity change sequence. Then, take the larger of the velocity sequence and the velocity change sequence point by point to form the upper velocity bound sequence. Perform amplitude limiting on each adjacent displacement to ensure that the adjacent displacement does not exceed the product of the corresponding upper velocity bound and the difference between adjacent timestamps to eliminate jump points. Output continuous positioning trajectory segments and write them to local storage.

5. A wearable wristwatch system for area control of workers according to claim 4, characterized in that: The location storage module further includes: For each written positioning trajectory segment, the byte sequence is concatenated in the order of timestamps. This byte sequence is then concatenated with the previously written check value, and a SHA256 one-way digest operation is performed to obtain the current check value, which is then written to local storage. At the same time, the timestamp and coordinates of the positioning trajectory segment are integerized and stored according to the adjacent difference, and the integerized step size is recorded as the upper bound of the error. This outputs a check chain that corresponds one-to-one with the positioning trajectory segment and is related to the previous and subsequent segments.

6. A wearable wristwatch system for area control of workers according to claim 5, characterized in that: The candidate event module includes: For each time stamp of the positioning trajectory segment, the signed distance to the boundary of the controlled area is calculated, and the median of the signed distances within the sliding time window is taken as the judgment distance. When the judgment distance changes from not less than zero to less than zero between adjacent time stamps, the next time stamp is recorded as the boundary start time stamp, and when the judgment distance changes from less than zero to not less than zero between adjacent time stamps, the next time stamp is recorded as the boundary end time stamp. Thus, boundary candidate events are output and written into the event time window determined by the boundary start time stamp and the boundary end time stamp. For the acquired sequence, the median of heart rate data and body temperature data are calculated as the baseline within the sliding time window, and the median of the absolute deviation is calculated as the fluctuation scale. At the same time, the resting segment is defined as the timestamp when the sum of acceleration data within the sliding time window is not greater than the median. When the deviation of heart rate data or body temperature data between adjacent timestamps in the resting segment changes from not greater than twice the fluctuation scale to greater than twice the fluctuation scale, it is recorded as the abnormal start timetamp. When the deviation changes from greater than twice the fluctuation scale to not greater than twice the fluctuation scale between adjacent timestamps, it is recorded as the abnormal end timetamp. Thus, abnormal candidate events are output and written into the event time window determined by the abnormal start timetamp and the abnormal end timetamp. Merge out-of-bounds candidate events and abnormal candidate events in chronological order, and concatenate events with an interval of less than the length of the sliding time window into a single event time window and write the event type, then output the candidate event set.

7. A wearable wristwatch system for area control of workers according to claim 6, characterized in that: The event logging module includes: For each candidate event in the candidate event set, extract the event trajectory from the positioning trajectory segment according to its event time window, and extract the event acquisition segment that matches the event time window timestamp from the acquisition sequence, and output the original data segment corresponding to the candidate event; The displacement of adjacent timestamp coordinates in the event trajectory is calculated and the velocity sequence is obtained by removing the bit and the difference between adjacent timestamps. The ratio of the maximum value to the median of the velocity sequence is taken as the trajectory continuity index. At the same time, the absolute difference between adjacent timestamps of skin contact data in the event acquisition segment is calculated and the median of the absolute difference is taken as the wearing stability index. The system calculates the resultant acceleration from the acceleration data in the event acquisition segment and takes the root mean square of the resultant acceleration within the event time window as the motion intensity index. The system also merges the trajectory continuity index, the wearing stability index, the motion intensity index, and the original data segment with the same candidate event identifier and writes them into the event record. The system outputs the event record that corresponds one-to-one with each candidate event.

8. A wearable wristwatch system for area control of workers according to claim 7, characterized in that: The verification module includes: Read the trajectory continuity index, wearing stability index and motion intensity index from the event log, multiply the trajectory continuity index and motion intensity index to obtain the upper limit value of motion, and take the reciprocal of the wearing stability index to obtain the wearing confidence value. Then, use the ratio of the upper limit value of motion to the wearing confidence value as the consistency value, and compare the consistency value with the fixed consistency threshold to output the alarm judgment result. When the alarm determination result is no alarm and the consistency value is within the warning range of the fixed consistency threshold, the sampling frequency of heart rate data and body temperature data is increased to twice the current sampling frequency. A new acquisition sequence and a new event record are generated with the new sampling frequency, and the consistency value is recalculated to update the alarm determination result.

9. A wearable wristwatch system for area control of workers according to claim 8, characterized in that: The verification module also includes: When the alarm determination result is no alarm and the consistency value is lower than the warning range, the event time window corresponding to the event record is extended forward and backward by a fixed extension duration. Based on this, the original data segment is re-extracted to generate a new event record. The consistency value is then recalculated to update the alarm determination result. The update stops when the consistency value remains unchanged for two consecutive calculations or when a sampling frequency increase and a time window extension have been performed once. The verified alarm event record is then output.

10. A wearable wristwatch system for area control of workers according to claim 9, characterized in that: The alarm reporting module includes: When the alarm determination result is an alarm, an alarm command is generated on the watch body and the sound, light or vibration output is driven. At the same time, the event type and event time window are extracted from the alarm event record, and the check value at the end of the check chain corresponding to the event time window is extracted to form a reporting packet and written into the local queue of reporting code. The reporting packet to be reported is output. The reported packets are numbered according to their generation sequence and reported to the management platform in numerical order when the network is available. If no reply is received from the platform, the reports are repeated at fixed retransmission intervals until a reply is received or the maximum number of retransmissions is reached. After receiving the reply, the reply number is written to local storage to form a closed-loop reporting record, thereby outputting an alarm record that can be verified and traced.

Images