An intelligent safety helmet based on internet of things and a positioning method thereof

Abstract

The present application relates to the technical field of Internet of Things, in particular to a kind of intelligent safety helmet based on Internet of Things and its positioning method, comprising the following steps: fusion air pressure and acceleration identification vertical carrying state, calibrate sensor zero offset, decoupling corrected acceleration component, identify travel and wake up positioning, calculate theoretical stop displacement, compare electronic fence output early warning information, in the present application, collect air pressure and acceleration utilize vertical carrying scene feature extraction zero offset parameter to carry out sensor self-calibration, eliminate the influence of cumulative error on motion attitude solution, accurately distinguish travel and in-place work state by calculating the dispersion degree of orthogonal axial component and active difference, realize positioning function on-demand wake-up, combine instantaneous ground speed and biological response mechanism to construct kinematic model, calculate braking distance and generate virtual vector end point, judge the risk of crossing the border in advance before touching the electronic fence, ensure that the work personnel have sufficient safety braking margin.

Description

Technical Field

[0001] This invention relates to the field of Internet of Things (IoT) technology, and in particular to an IoT-based smart safety helmet and its positioning method. Background Technology

[0002] The Internet of Things (IoT) technology refers to a network technology system that uses various information sensors, radio frequency identification (RFID) technology, global positioning systems (GPS), infrared sensors, laser scanners, and other devices and technologies to collect real-time information on any object or process that needs to be monitored, connected, or interacted with. This system collects various necessary information such as sound, light, heat, and electricity, and achieves ubiquitous connectivity between things and between things and people through various possible network access methods. It enables intelligent perception, identification, and management of objects and processes. Its core encompasses three levels: comprehensive perception, reliable transmission, and intelligent processing. Traditional smart safety helmets and their positioning methods involve installing a GPS chip or ultra-wideband (UWB) positioning tag inside or outside the helmet shell using mechanical connections such as buckles and screws. This is connected to a wireless communication circuit and a power supply battery. The positioning chip directly receives satellite signals to obtain latitude and longitude coordinates, or the positioning tag interacts with positioning base stations deployed at the construction site. The wireless communication circuit sends the acquired coordinate data or signal strength data to a remote server. The server marks points on an electronic map based on the latitude and longitude values, or calculates the wearer's planar coordinate position relative to the base station based on the signal arrival time difference and signal flight time.

[0003] Existing technologies rely on positioning chips or tags for continuous signal interaction. This all-time working mode ignores the wearer's actual active state of movement. Maintaining high-frequency communication during static or non-operational waiting periods results in ineffective power consumption. At the same time, existing electronic fence warning mechanisms only make static inclusion judgments based on the current instantaneous coordinates and boundaries, ignoring the inertial impulse and neural reaction delay during the movement of workers. It is difficult to reserve sufficient braking buffer distance in advance when approaching dangerous areas at high speed, resulting in personnel having already crossed the boundary when the safety warning is issued, causing substantial boundary crossing risks and safety management loopholes. Summary of the Invention

[0004] To achieve the above objectives, the present invention adopts the following technical solution: a smart safety helmet based on the Internet of Things and its positioning method, comprising the following steps: S1: Collect air pressure and acceleration data, perform differential calculations on air pressure data to obtain the vertical height change rate, calculate the dispersion of horizontal acceleration data to obtain the horizontal vibration intensity, and integrate and construct a basic motion feature set; S2: Based on the basic motion feature set, the vertical load state is determined by comparing the vertical height change rate with the elevator running speed reference range and the horizontal vibration intensity with the mechanical vibration static reference. The horizontal acceleration data is averaged to generate zero bias calibration parameters. S3: Based on the zero bias calibration parameters, the compensation acceleration data is used to obtain the corrected net acceleration and decompose it to obtain the vertical component set and the horizontal component set, and to construct the decoupled component data. S4: Calculate the discreteness of the vertical component set and the composite discreteness of the horizontal component set from the decoupled component data and perform differential analysis to obtain the activity difference result. Select the comparison result where the composite discreteness is greater than the travel intensity benchmark and the activity difference result is greater than the work posture judgment benchmark, output the positioning wake-up signal, and construct the positioning motion data. S5: Extract instantaneous ground speed and latitude and longitude from the positioning motion data, calculate the total theoretical stopping displacement by combining the reaction time constant and the human gait convergence deceleration factor, generate virtual vector end points by projecting latitude and longitude, substitute them into the electronic fence boundary to determine the inclusion relationship, and generate a safety warning signal.

[0005] As a further aspect of the present invention, the basic motion feature set includes a vertical pressure gradient sequence, horizontal acceleration frequency domain features, and multi-dimensional sensor time series labels. The zero-bias calibration parameters specifically include the inherent deviation value of the horizontal horizontal axis, the inherent deviation value of the horizontal vertical axis, and the zero-point drift temperature compensation coefficient. The decoupled component data specifically refers to the corrected triaxial resultant acceleration, the vertical projection vector of gravity direction, and the horizontal projection vector of the ground plane direction. The positioning motion data includes a satellite positioning activation status code, latitude and longitude coordinates, and Doppler frequency shift velocity measurement values. The safety warning signal specifically includes the audible and visual alarm trigger level, the virtual vector boundary crossing distance, and the electronic fence warning priority.

[0006] As a further aspect of the present invention, the step of obtaining the basic motion feature set specifically includes: S101: Using IoT sensors, it collects air pressure data and acceleration data, performs differential calculations on the continuously collected air pressure data, calculates the gradient change of air pressure values ​​within the continuous sampling period, converts the gradient change into vertical height variation values ​​based on the mapping relationship between air pressure and altitude, performs moving average processing and instantaneous fluctuation correction on the air pressure data sequence, and obtains the vertical height change rate. S102: Call the vertical height change rate, extract the horizontal axis acceleration data under the same time window according to the time series label, perform statistical variance calculation on the horizontal axis acceleration data, calculate the dispersion of the horizontal acceleration value relative to the mean, and obtain the horizontal vibration intensity. S103: Analyze the linear regression slope characteristics of the vertical height change rate over time, statistically analyze the numerical distribution characteristics of the horizontal vibration intensity in the time domain, align the vertical height change rate and horizontal vibration intensity data according to a unified time benchmark, associate and encapsulate the aligned slope characteristics and distribution characteristics, and establish a basic motion feature set.

[0007] As a further aspect of the present invention, the step of obtaining the zero-bias calibration parameter specifically includes: S201: Analyze and extract the vertical height change rate sequence and the corresponding horizontal vibration intensity sequence from the basic motion feature set, obtain the preset elevator running speed reference range and mechanical vibration static reference, compare the vertical height change rate with the elevator running speed reference range, compare the horizontal vibration intensity with the mechanical vibration static reference, integrate the comparison status of the two sets of data, and generate feature comparison logic results. S202: Call the feature comparison logic result, retrieve the time point where the vertical height change rate is within the velocity range and the horizontal vibration intensity is lower than the static reference, locate the start and end timestamps of the corresponding time segment, define the data interception range, and obtain the vertical transport time window; S203: Based on the vertical transport time window, backtrack and extract the original horizontal axial acceleration data within the target time range, construct a set of samples to be calibrated, and perform an arithmetic mean operation on the acceleration values ​​in the set as the inherent deviation benchmark of the sensor in the horizontal static state to generate zero bias calibration parameters.

[0008] As a further aspect of the present invention, the process of obtaining the preset elevator operating speed reference range and mechanical vibration static reference specifically includes: Historical vertical height change rate time series data and horizontal axial acceleration data are retrieved to obtain building single-story height parameters and elevator rated operating speed parameters. A division ratio operation is performed on the building single-story height parameters and the elevator rated operating speed parameters to generate a minimum single-trip operating cycle threshold. The vertical height change rate time series data is traversed using a sliding time window to identify continuous data segments with monotonically consistent numerical signs and durations exceeding the minimum single-trip operating cycle threshold. A simulated transport sample set is constructed, and the arithmetic mean of each segment in the simulated transport sample set is calculated. A vertical speed distribution dataset is constructed and Gaussian distribution fitting operation is performed. The expected value and standard deviation of the fitted curve are analyzed. The amplitude is expanded with the expected value as the center and three times the standard deviation as the boundary to construct the elevator operating speed benchmark range. Based on the timestamp index of the simulated vehicle sample set, horizontal acceleration segments are extracted from the horizontal axial acceleration data, the composite variance of the horizontal acceleration segments is calculated, the arithmetic mean and sample standard deviation of all the composite variances are calculated, and the arithmetic mean and three times the sample standard deviation are summed to obtain the mechanical vibration static reference.

[0009] As a further aspect of the present invention, the step of obtaining the decoupled component data specifically includes: S301: Call the zero-bias calibration parameters, synchronously collect the real-time acceleration data output by the sensor, perform subtraction compensation operation on the acceleration value at each discrete time point, convert the relative measurement values ​​of each axis into actual acceleration values ​​based on the absolute zero point, perform time-series arrangement and smoothing processing on the compensation results of multiple consecutive frames, and generate corrected net acceleration. S302: Based on the corrected net acceleration, a low-pass filter is used to extract the quasi-static DC component in the corrected net acceleration as the gravity direction vector, and a ground plane auxiliary vector perpendicular to the gravity direction is constructed. The corrected net acceleration is decomposed into vectors, and combined with the corresponding time series index, the vertical component set and the horizontal component set are obtained to generate the orthogonal axial component sequence. S303: Call the orthogonal axial component sequence, extract the vertical component set and the horizontal component set, combine with the corrected net acceleration, perform association mapping on the three sets of data according to the unified timestamp label, encapsulate and store the associated multidimensional data structure, and construct decoupled component data.

[0010] As a further aspect of the present invention, the step of acquiring the positioning motion data specifically includes: S401: Extract the vertical component set and the horizontal component set from the decoupled component data, perform variance statistical operation on the numerical sequence in the vertical component set to obtain vertical discrete values, calculate the combined vector magnitude sequence of the vector data in the horizontal component set and perform variance statistical operation to obtain horizontal composite discrete values, and obtain the activity difference result by performing difference operation on the two to generate a multidimensional discrete metric index. S402: Call the multidimensional discrete metric index to obtain the preset travel intensity benchmark and work posture judgment benchmark, filter the horizontal component set whose composite discreteness is greater than the travel intensity benchmark and whose activity difference result is greater than the comparison result of the work posture judgment benchmark, and generate a positioning wake-up signal. S403: In response to the positioning wake-up signal, obtain the current positioning timestamp, collect the instantaneous ground speed and latitude and longitude coordinates, and perform correlation and formatted encapsulation to construct positioning motion data.

[0011] As a further aspect of the present invention, the process of obtaining the preset travel intensity benchmark and the working posture determination benchmark specifically includes: Call historical decoupled component data, parse the vertical component set and the horizontal component set, calculate the vertical discrete value and the horizontal composite discrete value of each sampling point, sort the horizontal composite discrete value in ascending order, extract the first 50% of the data segments with the smallest value, and construct a silent calibration sample set. Calculate the arithmetic mean and sample standard deviation of the level synthetic discrete values ​​in the silent calibration sample set, and perform an addition operation on the arithmetic mean and 3 times the sample standard deviation to generate the marching intensity benchmark; The vertical discrete values ​​are extracted based on the timestamp index of the silent calibration sample set, and the numerical difference between the horizontal composite discrete value and the vertical discrete value corresponding to each sampling point is calculated to construct a differential distribution sequence.

[0012] Calculate the expected value and discrete deviation value of the differential distribution sequence, and perform an addition operation on the expected value and 3 times the discrete deviation value to obtain the work posture judgment benchmark.

[0013] As a further aspect of the present invention, the step of acquiring the safety warning signal specifically includes: S501: Extract the instantaneous ground velocity value from the positioning motion data, combine it with the preset reaction time constant and human gait convergence deceleration factor, calculate the buffer distance required for the target object to move from the current moment to a completely stationary state, and obtain the total theoretical stopping displacement; S502: Extract latitude and longitude coordinates from positioning motion data as the projection origin, analyze the vector direction of instantaneous ground speed, determine the heading angle, call the unit latitude and longitude distance conversion coefficient, convert the total theoretical stopping displacement into latitude increment values ​​and longitude increment values, perform numerical accumulation operation on latitude and longitude coordinates along the heading angle, and generate virtual vector end point; S503: Call the virtual vector endpoint, obtain the electronic fence boundary data, parse the polygonal geometric vertex sequence of the electronic fence area, substitute the virtual vector endpoint into the electronic fence boundary, compare the point coordinates with the geometric position relationship of the fence boundary, determine the inclusion relationship, and output a safety warning signal.

[0014] As a further aspect of the present invention, the process of obtaining the preset reaction time constant and braking deceleration constant specifically includes: The system retrieves a historical set of neural conduction response delay samples accumulated in the storage unit, performs normal distribution statistical operations on the neural conduction response delay sample set, calculates the arithmetic mean and discrete standard deviation of the sample population, performs a linear weighted summation operation on the arithmetic mean and the discrete standard deviation, configures the operation weight to 3, generates an upper limit value for the response delay distribution, and sets the upper limit value for the response delay distribution as the reaction time constant. Simultaneously, the database of historical workers' gait characteristics is accessed to extract the average number of steps and the gait frequency decay rate from receiving the instruction to stopping completely at different walking speeds. A gait convergence sample set is constructed, and linear regression analysis is performed on the gait convergence sample set to analyze the slope of the mapping relationship between speed and stopping distance, thereby generating the human gait convergence deceleration factor.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, air pressure and acceleration are collected to extract zero-bias parameters based on the characteristics of vertical transportation scenarios for sensor self-calibration, eliminating the influence of accumulated errors on motion attitude calculation. By calculating the dispersion and activity difference of orthogonal axial components, the invention accurately distinguishes between moving and stationary operation states, enabling on-demand activation of the positioning function. A kinematic model is constructed by combining instantaneous ground speed and biological reaction mechanisms to calculate the total theoretical stopping displacement, including braking distance, and generate a virtual vector endpoint. This allows for early prediction of boundary crossing risks before touching the electronic fence, ensuring that operators have sufficient safety braking margin. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the steps of the present invention; Figure 2 This is a detailed schematic diagram of S1 of the present invention; Figure 3 This is a detailed schematic diagram of S2 of the present invention; Figure 4 This is a detailed schematic diagram of S3 of the present invention; Figure 5 This is a detailed schematic diagram of S4 of the present invention; Figure 6 This is a detailed schematic diagram of S5 of the present invention. Detailed Implementation

[0018] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0019] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0020] Please see Figure 1 This invention provides an IoT-based smart safety helmet and its positioning method, comprising the following steps: S1: Collect air pressure and acceleration data, perform differential calculations on air pressure data to obtain the vertical height change rate, calculate the dispersion of horizontal acceleration data to obtain the horizontal vibration intensity, and integrate and construct a basic motion feature set; S2: Based on the basic motion feature set, the vertical load state is determined by comparing the vertical height change rate with the elevator running speed reference range and the horizontal vibration intensity with the mechanical vibration static reference. The horizontal acceleration data is averaged to generate zero bias calibration parameters. S3: Based on the zero bias calibration parameters, compensate for acceleration data to obtain corrected net acceleration and decompose it to obtain the vertical component set and the horizontal component set, and construct decoupled component data; S4: Calculate the discreteness of the vertical component set and the composite discreteness of the horizontal component set from the decoupled component data and perform differential analysis to obtain the activity difference results. Filter the results where the composite discreteness is greater than the travel intensity benchmark and the activity difference results are greater than the work posture judgment benchmark. Output the positioning wake-up signal and construct the positioning motion data. S5: Extract instantaneous ground speed and latitude and longitude from positioning motion data, calculate the total theoretical stopping displacement by combining reaction time constant and human gait convergence deceleration factor, generate virtual vector end points by projecting latitude and longitude, substitute them into the electronic fence boundary to determine the inclusion relationship, and generate a safety warning signal.

[0021] The basic motion feature set includes vertical pressure gradient sequence, horizontal acceleration frequency domain features, and multi-dimensional sensor time series labels. The zero-bias calibration parameters specifically include the inherent deviation value of the horizontal horizontal axis, the inherent deviation value of the horizontal vertical axis, and the zero-point drift temperature compensation coefficient. The decoupled component data specifically refers to the corrected triaxial resultant acceleration, the vertical projection vector of gravity direction, and the horizontal projection vector of the ground plane direction. The positioning motion data includes satellite positioning activation status code, latitude and longitude coordinates, and Doppler frequency shift velocity measurement values. The safety warning signals specifically include the audible and visual alarm trigger level, the virtual vector boundary crossing distance, and the electronic fence warning priority.

[0022] Please see Figure 2 The specific steps for obtaining the basic motion feature set are as follows: S101: Using IoT sensors, it collects air pressure data and acceleration data, performs differential calculations on the continuously collected air pressure data, calculates the gradient change of air pressure values ​​within the continuous sampling period, converts the gradient change into vertical height variation values ​​based on the mapping relationship between air pressure and altitude, performs moving average processing and instantaneous fluctuation correction on the air pressure data sequence, and obtains the vertical height change rate. By using a barometric pressure sensor and a six-axis inertial measurement unit (IMU) integrated inside the smart helmet, ambient air pressure and triaxial acceleration values ​​are simultaneously captured at a sampling frequency of 50 Hz, establishing a raw sensing data stream. After acquiring a continuous sequence of air pressure data, differential calculation logic is executed: the air pressure value at the current sampling moment is selected and subtracted from the air pressure value at the previous sampling moment to obtain the instantaneous air pressure fluctuation. A preset barometric altitude mapping coefficient is then used to multiply the instantaneous air pressure fluctuation by the mapping coefficient. For example, when the barometric altitude mapping coefficient is set to 0.09 meters per Pa, and the current air pressure is 101325 Pascals while the previous air pressure was 101326 Pascals, the difference between the two is -1 Pascal. According to the principles of atmospheric physics, air pressure decreases with increasing altitude; therefore, this negative difference corresponds to a vertical height change of 0.09 meters, representing an increase in altitude. Subsequently, to eliminate high-frequency noise caused by airflow disturbances, a moving average process was applied to the converted sequence of height variation values. A sliding time window containing five consecutive sampling points was constructed. The values ​​within the window were summed and divided by the window length to output the smoothed rate of change in vertical height. Assuming that the height variation values ​​of the five consecutive sampling points, after smoothing, the corrected rate of change in vertical height at that moment is 0.09 meters per sampling period, thus accurately quantifying the wearer's instantaneous displacement rate in the vertical dimension.

[0023] S102: Call the vertical height change rate, extract the horizontal axis acceleration data under the same time window based on the time series label, perform statistical variance calculation on the horizontal axis acceleration data, calculate the dispersion of the horizontal acceleration value relative to the mean, and obtain the horizontal vibration intensity. The vertical height change rate is invoked, and based on its time series label, horizontal axial acceleration data (X-axis and Y-axis acceleration readings) within the same time window is located and extracted from the raw acceleration data stream. For the extracted horizontal axial acceleration data, statistical variance calculation is performed to quantify the data dispersion. First, the arithmetic mean is obtained by summing and dividing all horizontal acceleration values ​​within the time window. Then, the difference between the acceleration value at each sampling point and this arithmetic mean is calculated, the difference is squared, and all squared terms are summed and divided by the total number of sampling points to obtain the horizontal vibration intensity. This indicator directly reflects the intensity of the wearer's head swaying in the horizontal plane, eliminating interference from gravitational components in constant-velocity motion or stationary states. For example, within a time window containing 10 sampling points, the mean horizontal acceleration is 0.2 meters per square second. If the fluctuation of the values ​​at each sampling point relative to the mean is large, the calculated variance is 0.05, and 0.05 is defined as the current horizontal vibration intensity. If the fluctuation is extremely small, the variance is 0.001, which indicates that the horizontal direction is relatively stable, providing data support for subsequent judgment on whether it is in a mechanically assisted lifting state.

[0024] S103: Analyze the linear regression slope characteristics of the vertical height change rate over time, statistically analyze the numerical distribution characteristics of the horizontal vibration intensity in the time domain, align the vertical height change rate and the horizontal vibration intensity data according to a unified time benchmark, associate and encapsulate the aligned slope characteristics and distribution characteristics, and establish a basic motion feature set. Linear regression analysis was performed on the evolution of the rate of change of vertical height over a continuous time axis. A specific time window (e.g., 2 seconds) was selected, and the data points of the rate of change of vertical height within this window were fitted using the least squares method. The slope of the fitted line was calculated. If the absolute value of the slope is close to zero, it indicates that the vertical rate is constant; if the slope is significant, it indicates that there is vertical acceleration or deceleration. Simultaneously, the numerical distribution of horizontal vibration intensity was statistically analyzed in the time domain, and the peak position and distribution width of its probability density function were calculated. Subsequently, a rigorous data alignment operation was performed, using a unified time base (e.g., the timestamp of Coordinated Universal Time, UTC) as an index, and the regression slope of the rate of change of vertical height was matched point by point with the horizontal vibration intensity sequence. The aligned slope features (reflecting vertical motion stability) and distribution features (reflecting horizontal vibration concentration) were correlated and encapsulated to construct a basic motion feature set containing multidimensional spatiotemporal information. In this feature set, for example, a data packet at a certain moment contains a combination of features: "vertical rise rate of 0.09 meters per sampling period" and "horizontal vibration variance of 0.001". This combination of features is marked and stored for subsequent accurate identification of specific non-autonomous motion scenarios such as riding a construction elevator, eliminating the risk of false alarms caused by single-dimensional data judgment.

[0025] Please see Figure 3 The specific steps for obtaining the zero-bias calibration parameters are as follows: S201: Analyze and extract the vertical height change rate sequence and the corresponding horizontal vibration intensity sequence from the basic motion feature set, obtain the preset elevator running speed reference range and mechanical vibration static reference, compare the vertical height change rate with the elevator running speed reference range, compare the horizontal vibration intensity with the mechanical vibration static reference, integrate the comparison status of the two sets of data, and generate the feature comparison logic result. The system retrieves preset elevator operating speed reference ranges (e.g., 1.5 m / s to 2.5 m / s) and mechanical vibration static reference ranges (e.g., variance threshold 0.02) from the storage unit. It then performs a dual logical comparison: first, it checks the real-time vertical height change rate against the elevator operating speed reference range; if the value falls within this range, the vertical state is marked as "pseudo-carrying"; simultaneously, it compares the corresponding horizontal vibration intensity with the mechanical vibration static reference range; if the vibration intensity is less than the static reference range, the horizontal state is marked as "relatively stationary". Finally, it integrates the two sets of comparison states to generate the feature comparison logic result. To establish the validity of the above benchmarks, historical data was pre-processed: Building single-story height parameters (e.g., 3 meters) and elevator rated operating speed parameters (e.g., 2 meters per second) were obtained, and a division operation was performed to obtain the minimum single-trip operating cycle threshold of 1.5 seconds. Historical vertical height change rates were iterated, and segments with durations exceeding 1.5 seconds and monotonically consistent numerical signs were identified to construct a suspected transport sample set. A Gaussian distribution was fitted to this set, yielding a mathematical expectation of 2.0 and a standard deviation of 0.1. Using the expectation value as the center and three times the standard deviation (0.3) as the radius, a benchmark range for elevator operating speeds of 1.7 to 2.3 meters per second was constructed. Similarly, horizontal acceleration segments within this time period were selected, and the combined variance was calculated. The mean was 0.005 and the standard deviation was 0.002. The summation (0.005 + 3 × 0.002) yielded 0.011, which was used as the mechanical vibration static benchmark.

[0026] S202: Call the feature comparison logic results, retrieve the time points where the vertical height change rate is within the velocity range and the horizontal vibration intensity is lower than the static reference, locate the start and end timestamps of the corresponding time segment, define the data interception range, and obtain the vertical transport time window; The feature comparison logic is invoked, and a continuous search is performed on the timeline to filter out time points that simultaneously meet the two conditions: "the rate of change of vertical height is within the elevator's operating speed reference range" and "the horizontal vibration intensity is lower than the mechanical vibration static reference." These consecutively satisfied time point sequences are identified, and the first satisfied time stamp is designated as the start time stamp, and the last satisfied time stamp as the end time stamp, thus defining a complete data extraction range and obtaining the vertical transport time series window. For example, if the search reveals that among 100 sampling points from time stamp T100 to T200, the vertical velocity remains consistently around 2.0 meters per second (within the range of 1.7-2.3), and the horizontal vibration intensity remains consistently at 0.008 (less than 0.011), then it is determined that the wearer is in a smoothly operating elevator during this time period, and T100 to T200 is defined as the vertical transport time series window. This precise window definition eliminates interference from people walking upstairs (large horizontal vibration) or jumping in place (discontinuous vertical velocity), locking in the optimal time for sensor zero-bias calibration.

[0027] S203: Based on the vertical transport time window, backtrack and extract the original horizontal axis acceleration data within the target time range, construct a set of samples to be calibrated, and perform an arithmetic mean operation on the acceleration values ​​in the set as the inherent deviation benchmark of the sensor in the horizontal static state to generate zero bias calibration parameters. Based on a defined vertical transport time window, horizontal axial acceleration data within this target time range are extracted from the original data records to construct a calibration sample set. Since the wearer is considered stationary relative to the elevator car within this time window, and the car experiences no significant horizontal acceleration, the output of the horizontal axial sensor should theoretically be zero. Therefore, the X-axis and Y-axis acceleration values ​​of all sampling points in this set are arithmetically averaged. The calculated average directly reflects the sensor's current zero-point drift and installation error. For example, in the 100 samples, the arithmetic mean of the X-axis acceleration readings is +0.03 m / s², and the Y-axis is -0.02 m / s². These two values ​​are then established as the zero-bias calibration parameters for the X-axis and Y-axis, respectively. This process utilizes the unique "static within motion" scenario of elevator operation to complete in-situ calibration of the sensor without manual intervention, providing a reliable correction benchmark for subsequent high-precision motion calculations.

[0028] Please see Figure 4 The specific steps for obtaining decoupled component data are as follows: S301: Call the zero-bias calibration parameters, synchronously collect the real-time acceleration data output by the sensor, perform subtraction compensation calculation on the acceleration value at each discrete time point, convert the relative measurement values ​​of each axis into actual acceleration values ​​based on the absolute zero point, perform time-series arrangement and smoothing processing on the compensation results of multiple consecutive frames, and generate corrected net acceleration. For each discrete time point, a subtraction compensation operation is performed on the original X-axis and Y-axis acceleration values. This involves subtracting the corresponding zero-bias calibration parameter from the original measurement value to restore the relative measurement values ​​along each axis to the actual acceleration values ​​based on the absolute physical zero point. For example, if the real-time X-axis acceleration is 0.05 m / s², and the zero-bias calibration parameter is 0.03 m / s², the compensated net acceleration is 0.02 m / s². To further suppress high-frequency random noise, the compensation results from multiple consecutive frames are sequentially arranged and smoothed using a low-pass filtering algorithm to generate a high signal-to-noise ratio corrected net acceleration sequence. This step effectively eliminates sensor errors caused by temperature drift or mechanical stress, ensuring that the input data for subsequent vector decomposition accurately reflects the wearer's kinematic characteristics.

[0029] S302: Based on the corrected net acceleration, a low-pass filter is used to extract the quasi-static DC component in the corrected net acceleration as the gravity direction vector, and a ground plane auxiliary vector perpendicular to the gravity direction is constructed. The corrected net acceleration is then decomposed into vectors, and combined with the corresponding time series index, the vertical component set and the horizontal component set are obtained to generate the orthogonal axial component sequence. By combining the principles of gravity sensing to determine the direction of gravity (i.e., the vertically downward direction, usually the negative Z-axis), the direction of the ground plane perpendicular to the direction of gravity is derived. Vector decomposition is performed on the three-axis corrected net acceleration vector, decomposing its projection into a set of vertical components along the direction of gravity and a set of horizontal components along the ground plane. Specifically, using trigonometric functions or quaternion rotation matrices, the acceleration vector in the sensor coordinate system is transformed to the geographic coordinate system, separating the pure vertical motion acceleration from the horizontal motion acceleration. For example, when the wearer is working with their head down, the sensor coordinate system tilts, and the gravity component originally on the Z-axis is dispersed onto the X and Y axes. Vector decomposition recorrects this, extracting the true horizontal travel acceleration component (e.g., 0.5 m / s²) and the residual acceleration after vertical gravity compensation (e.g., 0.1 m / s²), generating an orthogonal axial component sequence, thus decoupling the motion posture from the motion intensity.

[0030] S303: Call the orthogonal axial component sequence, extract the vertical component set and the horizontal component set, combine with the corrected net acceleration, perform correlation mapping on the three sets of data according to the unified timestamp label, encapsulate and store the correlation multidimensional data structure, and construct decoupled component data; The vertical and horizontal component sets are extracted and combined with the original corrected net acceleration data. These three sets of data are then mapped point-by-point using a unified high-precision timestamp label. A multi-dimensional data structure is constructed to package and encapsulate the "original corrected acceleration," "vertical component," and "horizontal component" at the same time, storing them in a decoupled component data buffer. For example, at timestamp T300, the encapsulated data packet contains the resultant acceleration with a vector magnitude of 9.81, a vertical undulation component with a value of 0.05, and a horizontal propulsion component with a vector magnitude of 0.8. This structured storage method not only preserves the original dynamic information but also provides component features after physical interpretation, providing a standardized input format for subsequent complex motion pattern recognition algorithms (such as gait analysis and fall detection), ensuring efficient data flow and logical closure.

[0031] Please see Figure 5 The specific steps for acquiring location motion data are as follows: S401: Extract the vertical component set and the horizontal component set from the decoupled component data, perform variance statistics on the numerical sequence in the vertical component set to obtain vertical discrete values, calculate the combined vector magnitude sequence of the vector data in the horizontal component set and perform variance statistics to obtain horizontal composite discrete values, and obtain the activity difference results by performing difference operations on the two to generate a multidimensional discrete metric index. Perform variance statistics on the numerical sequences within the vertical component set to quantify the vertical fluctuation energy (unit: m² / s). 4 The vertical discrete values ​​are obtained. Simultaneously, for the X-axis and Y-axis component data within the horizontal component set, the Pythagorean theorem is first used to calculate the resultant vector magnitude (i.e., the magnitude of horizontal acceleration, unit: m / s²) of each sampling point, constructing a magnitude sequence; then, variance statistics are performed on this magnitude sequence to obtain the horizontal composite discrete values ​​(unit: m² / s²). 4 This process ensures the consistency of physical dimensions between the horizontal and vertical indicators. Subsequently, a subtraction difference operation is performed on the horizontal composite discrete value and the vertical discrete value to obtain the activity difference result. This indicator aims to distinguish between "horizontal movement" and "stationary vertical operation" modes. For example, if the calculated horizontal composite discrete value is 0.15 and the vertical discrete value is 0.05, the activity difference result is 0.10. A positive and large value indicates that the movement is mainly concentrated in the horizontal direction, consistent with the characteristics of movement; conversely, if the difference result is negative, it indicates that vertical fluctuations are dominant. The generated multidimensional discrete metric constitutes the core basis for judging whether the wearer is in an effective movement state.

[0032] S402: Call the multidimensional discrete metric to obtain the preset travel intensity benchmark and work posture judgment benchmark, filter the horizontal component set whose composite discreteness is greater than the travel intensity benchmark and whose activity difference result is greater than the comparison result of the work posture judgment benchmark, and generate a positioning wake-up signal. The specific execution logic filtering process is as follows: Pre-calculated baseline parameters are retrieved from the storage unit, such as a travel intensity baseline set to 0.025 and a work posture determination baseline set to 0.028. The system monitors multi-dimensional discrete metrics in real time. When the system detects that the composite discreteness value of the horizontal component set at a certain moment is 0.08 (this value is greater than the travel intensity baseline of 0.025, indicating significant horizontal motion energy), and the calculated activity difference result is 0.05 (this value is greater than the work posture determination baseline of 0.028, indicating that the horizontal motion activity is significantly higher than the vertical activity, consistent with travel characteristics rather than stationary work), the system determines that the current state meets the positioning wake-up conditions and generates a positioning wake-up signal. This logic effectively filters out non-displacement actions such as squatting and standing in place or simple head rotation (the difference results for such actions are usually small or negative), ensuring that the high-power positioning module is activated only when travel actually occurs.

[0033] S403: Responds to the positioning wake-up signal, obtains the current positioning timestamp, collects instantaneous ground speed and latitude and longitude coordinates, and performs correlation and formatted encapsulation to construct positioning motion data; In response to the generated positioning wake-up signal, the device immediately wakes up from its dormant state to obtain the current positioning timestamp. It then collects instantaneous ground speed data (unit: meters per second) and high-precision latitude and longitude coordinates (format: degrees), and associates and formats this spatial location information with time information to construct positioning motion data. For example, at time T400 after receiving the wake-up signal, the instantaneous ground speed is collected as 1.5 meters per second, longitude as 116.397 degrees, and latitude as 39.908 degrees. This data is packaged, along with a current motion status tag (such as "in motion"), and stored in the transmission queue. This step ensures that trajectory points are recorded only when the wearer undergoes substantial displacement, avoiding positioning drift interference in a stationary state, while guaranteeing the timeliness and accuracy of the data, providing a solid physical coordinate foundation for subsequent safety warnings.

[0034] Please see Figure 6 The specific steps for obtaining safety warning signals are as follows: S501: Extract instantaneous ground velocity values ​​from positioning motion data, combine them with preset reaction time constants and human gait convergence deceleration factors, calculate the buffer distance required for the target object to move from the current moment to a completely stationary state, and obtain the total theoretical stopping displacement; The computational logic employs a biomechanical model based on human gait characteristics, which divides the stopping process into a neural reaction phase and a muscle-controlled deceleration phase. The system first extracts the instantaneous ground velocity and multiplies it by a preset reaction time constant to calculate the reaction displacement generated during the neural conduction lag. Then, it calls upon a human gait convergence deceleration factor, which characterizes the average deceleration capacity of the human body during a sudden stop by adjusting stride length and cadence. Based on the mapping relationship between the current instantaneous ground velocity and this deceleration factor, the system calculates the physical buffer distance required for the human body to come to a complete stop from the start of the braking action. Finally, the reaction displacement and the physical buffer distance are numerically summed to obtain the total theoretical stopping displacement. This value intuitively reflects the minimum safe space required to avoid exceeding the limits under the current movement speed and physiological function conditions.

[0035] S502: Extract latitude and longitude coordinates from positioning motion data as the projection origin, analyze the vector direction of instantaneous ground speed, determine the heading angle, call the unit latitude and longitude distance conversion coefficient, convert the total theoretical stopping displacement into longitude increment values ​​and latitude increment values, and generate virtual vector end points; First, the vector direction of the instantaneous ground velocity is analyzed to determine the wearer's heading angle. Given the non-linear characteristics of the geographic coordinate system, the system uses the current latitude coordinates and calls upon the unit latitude-longitude distance conversion coefficient in the geographic information system. This coefficient defines the physical distance on the ground corresponding to a unit of longitude and a unit of latitude at the current latitude. Using this conversion coefficient, the total theoretical stopping displacement, in meters, is vectorized and dimensionally transformed along the heading angle, calculating the corresponding longitude and latitude increments. Finally, these two increments are accumulated with the original positioning latitude and longitude coordinates to generate a virtual vector endpoint. This endpoint represents the extreme geographical location the wearer might ultimately reach due to inertia and reaction delay if a stop command were issued immediately.

[0036] S503: Call the virtual vector endpoint, obtain the electronic fence boundary data, parse the polygonal geometric vertex sequence of the electronic fence area, substitute the virtual vector endpoint into the electronic fence boundary, compare the point coordinates with the geometric position relationship of the fence boundary, determine the inclusion relationship, and output a safety warning signal; The method calls upon the virtual vector endpoint and retrieves the electronic fence boundary data from the safety management database, parsing the polygonal geometric vertex sequence of the area (such as a closed loop composed of a series of latitude and longitude coordinates). The virtual vector endpoint is then substituted into the electronic fence boundary model, and a ray casting algorithm or a roll-around algorithm is used to compare the geometric positional relationship between the endpoint coordinates and the fence boundary to determine inclusion. Specifically, it determines whether the endpoint is located inside the "no entry" area or outside the "safe operation" area defined by the electronic fence. If the determination result shows that the virtual vector endpoint has crossed the safety boundary (even if the current actual position has not yet crossed the boundary), a high-priority safety warning signal is immediately output. Table 1 shows the warning determination results at different speeds. Experimental data shows that, compared to the traditional triggering mechanism based solely on the current position, this method can issue an alarm approximately 1.5 seconds earlier when a person approaches the danger zone at a speed of 3 m / s, effectively avoiding substantial intrusion accidents caused by inertia.

[0037] Table 1. Example Data Table of Dynamic Electronic Fence Early Warning Judgment Test number Instantaneous ground speed (m / s) Current distance from the boundary (m) Total theoretical stopping displacement (m) Judgment result Alert status 001 1.0 5.0 0.83 Not crossed the boundary silence 002 3.0 3.0 3.25 Crossing the boundary Trigger alarm 003 2.0 2.5 1.91 Not crossed the boundary silence As shown in Table 1, in test number 002, although the current distance from the boundary is still 3.0 meters, the total theoretical stopping displacement reaches 3.25 meters due to the high speed, which exceeds the physical interval, thus triggering an alarm and verifying the scheme's ability to proactively perceive dynamic risks.

[0038] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A smart safety helmet based on the Internet of Things and its positioning method, characterized in that, Includes the following steps: S1: Collect air pressure and acceleration data, perform differential calculations on air pressure data to obtain the vertical height change rate, calculate the dispersion of horizontal acceleration data to obtain the horizontal vibration intensity, and integrate and construct a basic motion feature set; S2: Based on the basic motion feature set, the vertical load state is determined by comparing the vertical height change rate with the elevator running speed reference range and the horizontal vibration intensity with the mechanical vibration static reference. The horizontal acceleration data is averaged to generate zero bias calibration parameters. S3: Based on the zero bias calibration parameters, the compensation acceleration data is used to obtain the corrected net acceleration and decompose it to obtain the vertical component set and the horizontal component set, and to construct the decoupled component data. S4: Calculate the discreteness of the vertical component set and the composite discreteness of the horizontal component set from the decoupled component data and perform differential analysis to obtain the activity difference result. Select the comparison result where the composite discreteness is greater than the travel intensity benchmark and the activity difference result is greater than the work posture judgment benchmark, output the positioning wake-up signal, and construct the positioning motion data. S5: Extract instantaneous ground speed and latitude and longitude from the positioning motion data, calculate the total theoretical stopping displacement by combining the reaction time constant and the human gait convergence deceleration factor, generate virtual vector end points by projecting latitude and longitude, substitute them into the electronic fence boundary to determine the inclusion relationship, and generate a safety warning signal.

2. The smart safety helmet and its positioning method based on the Internet of Things according to claim 1, characterized in that, The basic motion feature set includes a vertical pressure gradient sequence, horizontal acceleration frequency domain features, and multi-dimensional sensor time series labels. The zero-bias calibration parameters specifically include the inherent deviation values ​​of the horizontal horizontal axis, the inherent deviation values ​​of the horizontal vertical axis, and the zero-point drift temperature compensation coefficient. The decoupled component data specifically refers to the corrected triaxial resultant acceleration, the vertical projection vector of gravity direction, and the horizontal projection vector of the ground plane direction. The positioning motion data includes satellite positioning activation status code, latitude and longitude coordinates, and Doppler frequency shift velocity measurement values. The safety warning signal specifically includes the audible and visual alarm trigger level, the virtual vector boundary crossing distance, and the electronic fence warning priority.

3. The smart safety helmet and its positioning method based on the Internet of Things according to claim 1, characterized in that, The specific steps for obtaining the basic motion feature set are as follows: S101: Using IoT sensors, it collects air pressure data and acceleration data, performs differential calculations on the continuously collected air pressure data, calculates the gradient change of air pressure values ​​within the continuous sampling period, converts the gradient change into vertical height variation values ​​based on the mapping relationship between air pressure and altitude, performs moving average processing and instantaneous fluctuation correction on the air pressure data sequence, and obtains the vertical height change rate. S102: Call the vertical height change rate, extract the horizontal axis acceleration data under the same time window according to the time series label, perform statistical variance calculation on the horizontal axis acceleration data, calculate the dispersion of the horizontal acceleration value relative to the mean, and obtain the horizontal vibration intensity. S103: Analyze the linear regression slope characteristics of the vertical height change rate over time, statistically analyze the numerical distribution characteristics of the horizontal vibration intensity in the time domain, align the vertical height change rate and horizontal vibration intensity data according to a unified time benchmark, associate and encapsulate the aligned slope characteristics and distribution characteristics, and establish a basic motion feature set.

4. The smart safety helmet and its positioning method based on the Internet of Things according to claim 3, characterized in that, The specific steps for obtaining the zero-bias calibration parameters are as follows: S201: Analyze and extract the vertical height change rate sequence and the corresponding horizontal vibration intensity sequence from the basic motion feature set, obtain the preset elevator running speed reference range and mechanical vibration static reference, compare the vertical height change rate with the elevator running speed reference range, compare the horizontal vibration intensity with the mechanical vibration static reference, integrate the comparison status of the two sets of data, and generate feature comparison logic results. S202: Call the feature comparison logic result, retrieve the time point where the vertical height change rate is within the velocity range and the horizontal vibration intensity is lower than the static reference, locate the start and end timestamps of the corresponding time segment, define the data interception range, and obtain the vertical transport time window; S203: Based on the vertical transport time window, backtrack and extract the original horizontal axial acceleration data within the target time range, construct a set of samples to be calibrated, and perform an arithmetic mean operation on the acceleration values ​​in the set as the inherent deviation benchmark of the sensor in the horizontal static state to generate zero bias calibration parameters.

5. The smart safety helmet and its positioning method based on the Internet of Things according to claim 4, characterized in that, The process of obtaining the preset elevator operating speed reference range and mechanical vibration static reference is as follows: Historical vertical height change rate time series data and horizontal axial acceleration data are retrieved to obtain building single-story height parameters and elevator rated operating speed parameters. A division ratio operation is performed on the building single-story height parameters and the elevator rated operating speed parameters to generate a minimum single-trip operating cycle threshold. The vertical height change rate time series data is traversed using a sliding time window to identify continuous data segments with monotonically consistent numerical signs and durations exceeding the minimum single-trip operating cycle threshold. A simulated transport sample set is constructed, and the arithmetic mean of each segment in the simulated transport sample set is calculated. A vertical speed distribution dataset is constructed and Gaussian distribution fitting operation is performed. The expected value and standard deviation of the fitted curve are analyzed. The amplitude is expanded with the expected value as the center and three times the standard deviation as the boundary to construct the elevator operating speed benchmark range. Based on the timestamp index of the simulated vehicle sample set, horizontal acceleration segments are extracted from the horizontal axial acceleration data, the composite variance of the horizontal acceleration segments is calculated, the arithmetic mean and sample standard deviation of all the composite variances are calculated, and the arithmetic mean and three times the sample standard deviation are summed to obtain the mechanical vibration static reference.

6. The smart safety helmet and its positioning method based on the Internet of Things according to claim 4, characterized in that, The specific steps for obtaining the decoupled component data are as follows: S301: Call the zero-bias calibration parameters, synchronously collect the real-time acceleration data output by the sensor, perform subtraction compensation operation on the acceleration value at each discrete time point, convert the relative measurement values ​​of each axis into actual acceleration values ​​based on the absolute zero point, perform time-series arrangement and smoothing processing on the compensation results of multiple consecutive frames, and generate corrected net acceleration. S302: Based on the corrected net acceleration, a low-pass filter is used to extract the quasi-static DC component in the corrected net acceleration as the gravity direction vector, and a ground plane auxiliary vector perpendicular to the gravity direction is constructed. The corrected net acceleration is decomposed into vectors, and combined with the corresponding time series index, the vertical component set and the horizontal component set are obtained to generate the orthogonal axial component sequence. S303: Call the orthogonal axial component sequence, extract the vertical component set and the horizontal component set, combine with the corrected net acceleration, perform association mapping on the three sets of data according to the unified timestamp label, encapsulate and store the associated multidimensional data structure, and construct decoupled component data.

7. The smart safety helmet and its positioning method based on the Internet of Things according to claim 6, characterized in that, The specific steps for acquiring the positioning motion data are as follows: S401: Extract the vertical component set and the horizontal component set from the decoupled component data, perform variance statistical operation on the numerical sequence in the vertical component set to obtain vertical discrete values, calculate the combined vector magnitude sequence of the vector data in the horizontal component set and perform variance statistical operation to obtain horizontal composite discrete values, and obtain the activity difference result by performing difference operation on the two to generate a multidimensional discrete metric index. S402: Call the multidimensional discrete metric index to obtain the preset travel intensity benchmark and work posture judgment benchmark, filter the horizontal component set whose composite discreteness is greater than the travel intensity benchmark and whose activity difference result is greater than the comparison result of the work posture judgment benchmark, and generate a positioning wake-up signal. S403: In response to the positioning wake-up signal, obtain the current positioning timestamp, collect the instantaneous ground speed and latitude and longitude coordinates, and perform correlation and formatted encapsulation to construct positioning motion data.

8. The smart safety helmet based on the Internet of Things and its positioning method according to claim 7, characterized in that, The process of obtaining the preset travel intensity benchmark and the work posture determination benchmark is as follows: Call historical decoupled component data, parse the vertical component set and the horizontal component set, calculate the vertical discrete value and the horizontal composite discrete value of each sampling point, sort the horizontal composite discrete value in ascending order, extract the first 50% of the data segments with the smallest value, and construct a silent calibration sample set. Calculate the arithmetic mean and sample standard deviation of the level synthetic discrete values ​​in the silent calibration sample set, and perform an addition operation on the arithmetic mean and 3 times the sample standard deviation to generate the marching intensity benchmark; The vertical discrete values ​​are extracted based on the timestamp index of the silent calibration sample set, and the numerical difference between the horizontal composite discrete value and the vertical discrete value corresponding to each sampling point is calculated to construct a differential distribution sequence. Calculate the expected value and discrete deviation value of the differential distribution sequence, and perform an addition operation on the expected value and 3 times the discrete deviation value to obtain the work posture judgment benchmark.

9. The smart safety helmet based on the Internet of Things and its positioning method according to claim 7, characterized in that, The specific steps for obtaining the security warning signal are as follows: S501: Extract the instantaneous ground velocity value from the positioning motion data, combine it with the preset reaction time constant and human gait convergence deceleration factor, calculate the buffer distance required for the target object to move from the current moment to a completely stationary state, and obtain the total theoretical stopping displacement; S502: Extract latitude and longitude coordinates from positioning motion data as the projection origin, analyze the vector direction of instantaneous ground speed, determine the heading angle, call the unit latitude and longitude distance conversion coefficient, convert the total theoretical stopping displacement into latitude increment values ​​and longitude increment values, perform numerical accumulation operation on latitude and longitude coordinates along the heading angle, and generate virtual vector end point; S503: Call the virtual vector endpoint, obtain the electronic fence boundary data, parse the polygonal geometric vertex sequence of the electronic fence area, substitute the virtual vector endpoint into the electronic fence boundary, compare the point coordinates with the geometric position relationship of the fence boundary, determine the inclusion relationship, and output a safety warning signal.

10. The smart safety helmet and its positioning method based on the Internet of Things according to claim 9, characterized in that, The process of obtaining the preset reaction time constant and braking deceleration constant specifically includes: The system retrieves a historical set of neural conduction response delay samples accumulated in the storage unit, performs normal distribution statistical operations on the neural conduction response delay sample set, calculates the arithmetic mean and discrete standard deviation of the sample population, performs a linear weighted summation operation on the arithmetic mean and the discrete standard deviation, configures the operation weight to 3, generates an upper limit value for the response delay distribution, and sets the upper limit value for the response delay distribution as the reaction time constant. Simultaneously, the database of historical workers' gait characteristics is accessed to extract the average number of steps and the gait frequency decay rate from receiving the instruction to stopping completely at different walking speeds. A gait convergence sample set is constructed, and linear regression analysis is performed on the gait convergence sample set to analyze the slope of the mapping relationship between speed and stopping distance, thereby generating the human gait convergence deceleration factor.

Images