Human factor intelligent-based ergonomics risk monitoring method, edge computing device, system and medium

Abstract

The embodiment of the application provides a kind of based on ergonomics risk monitoring method, edge computing device, system and medium of human factor intelligence, wherein, method includes: by action capture device and electromyography equipment respectively at least two body parts of data acquisition object in the process of executing task Action data and electromyography data are acquired;Based on action data, the basic risk assessment index of each body part of data acquisition object is determined;According to the electromyography data of each body part acquired at the same time, the weight of the basic risk assessment index of each body part is calibrated;Based on the basic risk assessment index of each body part after calibration and corresponding weight, the ergonomics risk monitoring result of data acquisition object is determined.The technical scheme provided in the embodiment of the application effectively improves the accuracy of ergonomics risk assessment by electromyography data.

Description

Technical Field

[0001] This application relates to the fields of human-machine ergonomics assessment and industrial engineering technology, and in particular to an ergonomics risk monitoring method, edge computing device, system and medium based on human factors intelligence. Background Technology

[0002] Currently, the methods for monitoring work ergonomic risks in assembly scenarios can score the working postures of assembly workers through manual observation or offline video. Although this method can identify high-risk postures to a certain extent, it suffers from poor real-time performance, strong subjectivity, and difficulty in quantifying muscle load and fatigue accumulation, which means that the accuracy of work ergonomic risk assessment needs to be further improved. Summary of the Invention

[0003] This application provides a human factors intelligence-based ergonomics risk monitoring method, edge computing device, system, and medium to improve the accuracy of ergonomics risk assessment.

[0004] In a first aspect, embodiments of this application provide a method for monitoring work-related risks based on human factors intelligence, comprising: acquiring motion data and electromyographic data of at least two body parts of a data acquisition subject during task execution using a motion capture device and an electromyographic device, respectively; determining basic risk assessment indicators for each body part of the data acquisition subject based on the motion data; calibrating the weights of the basic risk assessment indicators for each body part based on the electromyographic data of each body part acquired simultaneously; and determining the work-related risk monitoring results of the data acquisition subject based on the calibrated basic risk assessment indicators and corresponding weights for each body part.

[0005] In one possible implementation, the above-mentioned acquisition of motion data and electromyographic (EMG) data of at least two body parts of the data acquisition subject during task execution via motion capture equipment and EMG equipment includes: acquiring first motion data, second motion data, and corresponding EMG data of the data acquisition subject during task execution; the first motion data refers to the motion data of a first body part of the data acquisition subject, and the second motion data refers to the motion data of a second body part of the data acquisition subject; the first body part includes a first muscle group and a second muscle group, and the second body part includes a third muscle group and a fourth muscle group; the EMG data includes the EMG data of the first muscle group, the second muscle group, the third muscle group, and the fourth muscle group; and determining basic risk assessment indicators for each body part of the data acquisition subject based on the motion data includes: determining a first basic risk assessment indicator for the first body part based on the first motion data, and a second basic risk assessment indicator based on the second motion data. Motion data is used to determine a second basic risk assessment index for a second body part. The weights of the basic risk assessment indices for each body part are calibrated based on electromyographic (EMG) data acquired simultaneously from each body part, including: determining a first muscle group load index based on EMG data from a first muscle group and a second muscle group; determining a second muscle group load index based on EMG data from a third and fourth muscle group; calibrating a first weight based on the first muscle group load index, where the first weight is the weight of the first basic risk assessment index; calibrating a second weight based on the second muscle group load index, where the second weight is the weight of the second basic risk assessment index; determining the overall risk level of the data collection object during task execution based on the first basic risk assessment index, the second basic risk assessment index, and the calibrated first and second weights; and determining the ergonomic risk monitoring results of the data collection object during task execution based on the overall risk level.

[0006] In one possible implementation, the above-mentioned calibration of the first weight based on the first muscle group load index includes: Based on the first basic risk assessment indicator, the first local risk level is determined, which is the risk level of the first part of the body. Based on the first muscle group load index and the first local risk level, a first load residual is determined, which is the load residual of the first part of the body; When the value of the first load residual is positive, the first weight is increased; when the value of the first load residual is negative, the first weight is decreased.

[0007] In one possible implementation, before calibrating the first weight based on the first muscle group load index, the following steps are also included: The risk assessment index for the use of the first muscle is determined based on the load index of the first muscle group. The risk assessment index for the use of the first muscle is the risk assessment index for the use of the first muscle in the body. Based on the first motion data, a first static hold risk assessment index and a first load / force risk assessment index are determined. The first static hold risk assessment index is the static hold risk assessment index of the first part of the body, and the first load / force risk assessment index is the load / force risk assessment index of the first part of the body. Based on the first muscle group load index, the first weight is calibrated, including: If the conditions for the first muscle use risk assessment index, the first static hold risk assessment index, and the first load / force risk assessment index are all not met, the first weight is calibrated based on the first muscle group load index.

[0008] In one possible implementation, the first weight remains unchanged if any one of the first muscle use risk assessment index, the first static holding risk assessment index, and the first load / force risk assessment index meets the conditions. Accordingly, before determining the overall risk level of the data collection object during task execution based on the first basic risk assessment indicator, the second basic risk assessment indicator, and the calibrated first and second weights, the following steps are also included: The third weight is calibrated. The third weight is the weight of the third indicator. The third indicator is any one of the following indicators that meets the conditions: the first muscle use risk assessment indicator, the first static retention risk assessment indicator, and the first load / force risk assessment indicator. Accordingly, based on the first basic risk assessment indicator, the second basic risk assessment indicator, and the calibrated first and second weights, the overall risk level of the data collection object during the task execution process is determined, including: Based on the first basic risk assessment indicator, the second basic risk assessment indicator, the first weight, the calibrated second weight, the third indicator, and the calibrated third weight, the overall risk level of the data collection object during the task execution process is determined.

[0009] In one possible implementation, the above-mentioned calibration of the third weight includes: Based on the first basic risk assessment indicator and the third indicator, the first local risk level is determined, which is the risk level of the first part of the body. Based on the first muscle group load index and the first local risk level, the first load residual is determined, which is the load residual of the first part of the body. When the value of the first load residual is positive, the third weight is increased; when the value of the first load residual is negative, the third weight is decreased.

[0010] In one possible implementation, the first weight remains unchanged if any two of the first muscle use risk assessment index, the first static holding risk assessment index, and the first load / force risk assessment index meet the conditions. Accordingly, before determining the overall risk level of the data collection object during task execution based on the first basic risk assessment indicator, the second basic risk assessment indicator, and the calibrated first and second weights, the following steps are also included: Determine the first load impact index and the second load impact index. The first load impact index is the impact index of the fourth index on the load, and the second load impact index is the impact index of the fifth index on the load. The fourth index and the fifth index can be any two indices. When the first load impact index is greater than the second load impact index, the fourth weight is calibrated, and the fourth weight is the weight of the fourth index. Accordingly, based on the first basic risk assessment indicator, the second basic risk assessment indicator, and the calibrated first and second weights, the overall risk level of the data collection object during the task execution process is determined, including: Based on the first basic risk assessment indicator, the second basic risk assessment indicator, the first weight, the calibrated second weight, the fourth indicator, the fifth indicator, the fifth weight, and the calibrated fourth weight, the overall risk level of the data collection object in the process of performing the task is determined, and the fifth weight is the weight of the fifth indicator.

[0011] In one possible implementation, determining the first load impact index and the second load impact index includes: Based on the load residuals before and after the change of the fourth indicator, the first load impact indicator is determined. Based on the load residuals before and after the change of the fifth indicator, the second load impact indicator is determined.

[0012] In one possible implementation, the aforementioned first muscle group uses risk assessment indicators that meet certain conditions, including: the first muscle group load indicator is a high load indicator, or the proportion of the first muscle group load indicator as a high load indicator within a preset time period reaches a first threshold, and the high load indicator is a first muscle group load indicator that exceeds a second threshold.

[0013] In one possible implementation, before determining the efficiency risk of the data collection object during task execution based on the overall risk level, the following steps are also included: Based on the confidence level of the data acquisition channels, a sixth weight is determined. The sixth weight is the weight of the overall risk level. The data acquisition channels include the acquisition channels for motion data and the acquisition channels for electromyography data. Accordingly, the above-mentioned determination of the efficiency risk of data collection objects during task execution based on the overall risk level includes: determining efficiency risk based on the overall risk level and the sixth weight; or, Based on the overall risk level, the above-mentioned efficiency risks of data collection objects during task execution are determined, including: Based on the load indicators of the first muscle group, the load indicators of the second muscle group, and the overall risk level, determine the ergonomic risk; or... After determining the efficiency risks of data collection targets during task execution based on the overall risk level, the following also includes: Based on efficiency risks, reverse optimization is performed to generate improvement suggestions.

[0014] In one possible implementation, the determination of the load index of the first muscle group based on the electromyographic data of the first muscle group and the electromyographic data of the second muscle group includes: Extract the first temporal intensity feature from the electromyography data of the first muscle group, and extract the second temporal intensity feature from the electromyography data of the second muscle group; The intensity features in the first and second time domains were standardized to obtain normalized electromyographic data for the first and second muscle groups. Based on the normalized electromyographic data of the first and second muscle groups, the load index of the first muscle group was determined.

[0015] In a second aspect, embodiments of this application provide an edge computing device, including: one or more processors and a memory, wherein the one or more processors are coupled to the memory, the memory is used to store a computer program, and when the one or more processors execute the computer program, the edge computing device performs the method as described in any of the first aspects above.

[0016] Thirdly, embodiments of this application provide an ergonomic risk monitoring system based on human factors intelligence, including a main control terminal, a motion capture unit, and an electromyography (EMG) acquisition unit. The main control terminal is connected to the motion capture unit and the EMG acquisition unit, respectively. The motion capture unit is used to collect motion data, the EMG acquisition unit is used to collect EMG data, and the main control terminal is used to execute the method described in any of the first aspects above.

[0017] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, wherein when the computer program is run on a computer, the computer executes the method described in any of the first aspects above.

[0018] The human factors intelligence-based ergonomic risk monitoring method, edge computing device, system, and medium provided in this application improve the consistency between risk assessment and actual muscle load by acquiring electromyographic data from various parts of the body and calibrating the weights of basic risk assessment indicators based on the electromyographic data, thereby effectively improving the accuracy of ergonomic risk assessment. Attached Figure Description

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

[0020] Figure 1 A schematic diagram of the structure of an ergonomic risk monitoring system based on human factors intelligence provided in this application embodiment; Figure 2 This application provides a schematic diagram of the structure of a master control / edge computing device. Figure 3 A schematic diagram of another human factors intelligence-based ergonomic risk monitoring system provided in this application embodiment; Figure 4 A flowchart illustrating a human factors intelligence-based ergonomic risk monitoring method provided in this application embodiment; Figure 5 A flowchart illustrating another human factors intelligence-based ergonomic risk monitoring method provided in this application embodiment; Figure 6 A schematic diagram of a human body thermogram provided in an embodiment of this application; Figure 7 A schematic diagram illustrating the simulation comparison of the change of the human ergonomic risk index over time before and after optimization, provided for an embodiment of this application; Figure 8 A simulation comparison diagram of the estimated reduction before and after optimization is provided for an embodiment of this application; Figure 9 This is a schematic diagram of the structure of an edge computing device provided in an embodiment of this application. Detailed Implementation

[0021] The terminology used in the implementation section of this application is for the purpose of explaining specific embodiments of this application only, and is not intended to limit this application.

[0022] In this application's embodiments, ergonomic risk refers to human ergonomic risk.

[0023] The ergonomic risk monitoring method provided in this application embodiment can be applied to the main control terminal of an ergonomic risk monitoring system.

[0024] like Figure 1 As shown, the ergonomic risk monitoring system may include a motion capture (MoCap) unit, an electromyography (EMG) acquisition unit, and a main control terminal. Both the motion capture unit and the EMG acquisition unit are connected to the main control terminal.

[0025] The motion capture unit can be an inertial measurement unit (IMU), such as a 6-axis IMU containing accelerometers and gyroscopes, or a 9-axis IMU containing accelerometers, gyroscopes, and magnetometers. The IMU can be worn or attached to key joints of the torso, shoulders, elbows, wrists, thighs, and calves of workers in work scenarios (such as assembly, handling, picking, and care). During the worker's work, the IMU at each joint outputs the linear acceleration and angular velocity of that joint. For example, the IMU on the torso outputs the linear acceleration and angular velocity of that torso, the IMU on the elbow outputs the linear acceleration and angular velocity of that torso, and so on. The main control unit can perform Kalman filtering or Madgwick algorithm to fuse the pose quaternions from the linear acceleration and angular velocity of each joint, reconstructing the Euler angles of each joint and the skeletal pose.

[0026] The motion capture unit can also be an optical motion capture device, which may include marker light spots, a light spot capture unit, and corresponding light spot information processing software. Multiple marker light spots can be placed at various joint positions of the worker, forming a light spot array. The optical capture unit collects the light spot array information and outputs it to the main control terminal. The light spot information processing software can be installed on the main control terminal, identifying the worker's skeletal marker points based on the reflective light spot array, obtaining the worker's joint positions through a 3D spatial reconstruction algorithm, and calculating the worker's relative angles and postures.

[0027] Alternatively, the motion capture unit may include either an inertial measurement unit (IMU) for providing continuous pose estimation or an optical motion capture device for providing a spatial reference for position and pose, enabling fusion compensation and redundancy tolerance. In this case, the motion capture unit may also include a fusion module for outputting joint angle sequences (such as shoulder flexion / extension, elbow flexion / extension, wrist dorsiflexion, etc.), pose quaternions / Euler angles, angular velocity, and linear acceleration.

[0028] The electromyography (EMG) acquisition unit can be a multi-channel wireless surface electromyography (sEMG) device. Electrodes are arranged for typical muscle groups under load based on the key movements involved in the process, such as the upper / middle trapezius, anterior deltoid, biceps brachii, and triceps brachii in the upper limbs, and the erector spinae and quadratus lumborum in the trunk. Each muscle group can be configured with three electrodes (one of which is a reference electrode), forming an EMG data acquisition channel to acquire EMG data for one muscle group. For example, electrodes can be applied along the center of the muscle belly. When the worker is standing during the operation, electrodes can also be placed on the quadriceps femoris, biceps femoris, and gastrocnemius muscles in the lower limbs to acquire EMG data for these muscle groups. For example, the acquisition path can be completed using an electrode layout diagram and a channel mapping table. Completing the acquisition path means knowing which channel acquires the EMG data for which muscle group.

[0029] The control unit can be an edge computing device, such as an Industrial Personal Computer (IPC) or a Personal Computer (PC).

[0030] like Figure 2 As shown, the main control terminal may include a main control host and a display terminal connected to the main control host. The main control host includes a hardware interface layer and a software layer that runs in the hardware.

[0031] The hardware interface layer includes a multimodal synchronization card, communication interfaces, and a storage array. The multimodal synchronization card can be a transistor-transistor logic input / output (TTL I / O), a general-purpose input / output (GPIO), or a Labstreaming Layer (LSL Hub), etc. The multimodal synchronization card is used for hardware alignment; for example, it can convert synchronization commands from the master control unit into hardware trigger signals (such as TTL pulses) that can be recognized by each acquisition device, and simultaneously distribute them to the IMU, EMG, and optical system, thereby allowing all devices to sample at the same time. The communication interface can be a Universal Serial Bus (USB), a hub, an Ethernet switch, Bluetooth, or Wireless Fidelity (WiFi), etc. The storage array can be a Solid State Disk (SSD) or a Redundant Array of Independent Disks (RAID), etc., used to cache the data received from each acquisition unit by the communication interface to avoid packet loss due to bus delay or excessive instantaneous data volume.

[0032] The software layer can provide core services such as acquisition management, synchronization management, preprocessing, RULA engine, EMG fusion, ERI calculation, simulation adapter, optimizer, visualization / reporting, etc.

[0033] Specifically, the software layer may include a synchronization management module, an acquisition management module, a data processing module, a Rapid Upper Limb Assessment (RULA) assessment engine, an electromyography (EMG) fusion module, an Efficiency Risk Fusion Index (ERI) calculation and risk assessment module, a simulation adapter, an optimizer, a visualization and reporting module, and an interface architecture module to provide the aforementioned core services.

[0034] The synchronization management unit is used for software time base management. It can generate synchronization pulses and send them to the acquisition unit through the communication interface.

[0035] The acquisition and management module can retrieve the connected data streams from the storage array at the hardware interface layer, regardless of whether these data were initially transmitted via hardware interfaces such as USB / TCP / serial port or via LSL. LSL is a data stream protocol used to unify the access of data from different devices to the hardware interface layer. Furthermore, the acquisition and management module can also be used to uniformly manage the access, buffer reading, and basic format standardization of various data streams, and send the data to the quality control module for integrity detection such as noise, drift, and packet loss.

[0036] The data processing module reads or receives data from the acquisition and management module, and performs quality monitoring, noise reduction, feature extraction (such as joint angle and muscle group load) on the read or received data (such as motion data and electromyography data), and generates the preprocessing results required for subsequent scoring and fusion calculation.

[0037] The EMG fusion module is used for posture-electromyography consistency calibration, such as calibrating the weights of basic risk assessment indicators, muscle use risk assessment indicators, static hold risk assessment indicators, and load / force risk assessment indicators based on muscle group load indicators.

[0038] The RULA assessment engine is used to obtain basic risk assessment indicators based on the processing results of the data processing module, such as determining the basic risk assessment indicators for each body part based on the joint angles calculated by the data processing module.

[0039] The ERI calculation and risk assessment module is used to calculate the ERI / cumulative ergonomic risk (CPL) based on the processing results of the EMG fusion module. CPL is explained in detail below.

[0040] The simulation adapter can be an Application Programming Interface (API) used to provide a virtual simulation platform (such as Jack). ® ), CATIA ® The simulation adapter provides an interface for various functions, including: simulation of joint torques, accessibility, field of view, and collision. Simulation verification (joint torques, accessibility, field of view, collision) can be performed in an external virtual simulation platform (such as Jack, CATIA, etc.). The simulation adapter serves as the interface for transmitting and receiving results.

[0041] The optimizer is used to optimize a work scenario or task and generate an optimized solution. The optimization algorithm used by the optimizer can be non-dominated sorting genetic algorithm II, simulated annealing (SA), Bayesian optimization (BO), etc.

[0042] The visualization and reporting module can be used to generate display information such as alarms, heatmaps, and time curves, and can also generate a web visualization interface for display terminals. Specifically, the visualization and reporting module can perform risk hotspot overlay (body anatomy / virtual human model surface coloring), risk timeline / peak event tracking, automatic generation of parameterized suggestions (table height / clamp angle / tool ​​suspension / process position, etc.), report export, and API interface, etc.

[0043] The interface-based architecture module can send optimization plans to enterprise system interfaces such as Manufacturing Execution System (MES), Industrial Engineering (IE), and Product Lifecycle Management API (PLM API). Specifically, it can write back structured data such as optimization parameters, KPI indicators, risk reduction, and report indexes through the API.

[0044] In some embodiments, the master control terminal may include a ratio Figure 2 The main control unit may have more or fewer modules. For example, the main control unit may omit the display terminal, or the emulation adapter, or the optimizer, and so on.

[0045] In some embodiments, the ergonomic risk monitoring system may also include, for example: Figure 3 As shown, it further includes an event recording unit connected to the main control terminal. For example, the event recording unit can be a camera that acquires images of the work scene, including the operator. The main control terminal can recognize the work scene images to determine any one or more of the following: process start / end, key actions, and workload changes. Key actions can be customized according to the industry scenario. For example, key actions in an assembly scenario include hand movements (such as grasping, holding, screw rotation, insertion / removal, flipping, etc.), upper limb movements (such as lifting, lowering, abduction, adduction, etc.), and torso movements (such as bending, twisting, etc.). Tool load changes may include using different load tools (such as manual / electric wrenches, welding torches, etc.), changes in workpiece center of gravity (such as lifting a 3kg workpiece and then becoming unloaded, etc.), and switching of tooling attachment structures (such as auxiliary support rotation / stretching actions, etc.). The main control terminal can also record the time points and context of key process events based on the judgment results.

[0046] In some embodiments, the event recording unit can also be a foot pedal or handheld marker. The operator can trigger event signals (such as "assembly start", "tool contact", "load change" etc.) through the buttons on the marker. The main control terminal receives the event signals and writes them synchronously into the system timeline, thereby ensuring that all collected data can be aligned under the same time reference.

[0047] The human factors intelligence-based ergonomic risk monitoring method provided in this application improves the accuracy of ergonomic risk assessment by acquiring electromyographic data while acquiring motion data, and calibrating the weights of basic risk assessment indicators for different body parts of workers based on the electromyographic data.

[0048] Taking assembly scenarios as an example, the ergonomic risk monitoring method based on human factors intelligence provided in this application embodiment can be as follows: Figure 4 As shown, it includes: Step 401: Acquire motion data and electromyographic data of at least two body parts of the data acquisition subject during the performance of the task using motion capture equipment and electromyography equipment.

[0049] For example, during the operation of the worker, the main control unit acquires motion data from the motion capture unit and electromyography data from the electromyography acquisition unit.

[0050] Step 402: Based on motion data, determine the basic risk assessment indicators for each body part of the data collection subject. For example, the main control unit determines the basic risk assessment indicators for each body part of the worker based on motion data.

[0051] After acquiring motion data from the motion capture unit, the main control unit can identify the corresponding body parts: upper arm, forearm, wrist, neck, torso, and legs. Based on the corresponding motion data, it can calculate the joint angle range and posture maintenance duration of these parts, and use this to determine the basic risk assessment indicators.

[0052] For example, the main control unit calculates the elbow joint angle α based on the upper arm's motion data. When α ≤ 20°, the basic risk assessment index of the upper arm = 1; when 20° < α ≤ 45°, the basic risk assessment index of the upper arm = 2; when 45° < α ≤ 90°, the basic risk assessment index of the upper arm = 3; when α > 90°, the basic risk assessment index of the upper arm = 4; when it is determined from the upper arm's motion data that the upper arm also has abduction / shrugging / crossing the midline, the basic risk assessment index of the arm is increased by 1.

[0053] The master control unit calculates the relevant angle β based on the forearm's motion data. When 60° < β ≤ 100°, the basic risk assessment index of the forearm = 1; when β is other angles, the basic risk assessment index of the forearm = 2.

[0054] The main control unit calculates the wrist flexion angle based on the wrist movement data. When the flexion angle δ of the back of the hand / palm is ≤15°, the basic risk assessment index of the wrist is 1; when δ>15°, the basic risk assessment index of the wrist is 2; when the radial / ulnar deviation is obvious based on the wrist movement data, such as when the direction of the flexion angle is obviously towards the inner or outer side of the body, the basic risk assessment index of the wrist is increased by 1.

[0055] The main control unit calculates the neck torsion angle ε based on the neck movement data. When 0 < ε ≤ 10°, the basic risk assessment index of the neck = 1; when 10° < ε ≤ 20°, the basic risk assessment index of the neck = 2; when ε > 20°, the basic risk assessment index of the neck = 3; when it is determined from the neck movement data that the neck is significantly tilted backward or rotated to the side, if the direction of the neck torsion angle is obviously backward or to one side, the basic risk assessment index of the neck is increased by 1.

[0056] The main control unit calculates the torso's torso twisting angle φ based on the torso's motion data. When 0 < φ ≤ 5°, the basic risk assessment index of the torso = 1; when 5° < φ ≤ 20°, the basic risk assessment index of the torso = 2; when 20° < φ ≤ 60°, the basic risk assessment index of the torso = 3; when φ > 60°, the basic risk assessment index of the torso = 4; when the torso's motion data indicates that the torso is leaning backward or twisting significantly, if the direction of the torso's torso twisting angle is clearly backward or to one side, the basic risk assessment index of the torso is increased by 1.

[0057] The master control unit determines whether the legs are stably supported / in good sitting posture based on leg motion data. If the legs are stably supported / in good sitting posture, the basic risk assessment index for the legs is 1; if they are not stably supported, the basic risk assessment index for the legs is 2. The conditions for stable leg support are: a double-foot contact ratio ≥ 80%, sufficient support base, and stable pelvic projection. The master control unit can determine whether the distance between the height of the soles / heels of both feet and the contact height reference is within ±2cm based on leg motion data. If so, double-foot contact is determined. The double-foot contact ratio refers to the proportion of the number of times the feet make contact within a predetermined time (e.g., 3 seconds) to the total number of times the feet make contact (the sum of the number of times the feet make contact and the number of times the feet do not make contact). The contact height reference can be determined before the operator performs the task. For example, the operator stands still for 3 seconds before performing the task to allow the master control unit to estimate the contact height reference between the ground plane and each foot. The estimation can be performed by processing the data signal in frames using a 2-second sliding window and a 0.1-second step size. A method to determine if the support base is sufficient can be: calculate the horizontal distance between the ankles based on leg movement data. If the horizontal distance between the ankles is ≥15 cm (or ≥0.6 times the foot length), then the support base is considered sufficient (i.e., wide enough). A method to determine if the pelvic projection is stable can be: calculate the standard deviation of the pelvic horizontal displacement based on leg movement data. If the standard deviation of the pelvic horizontal displacement is <3 cm, and the pelvic projection falls within the support polygon formed by the line connecting the two feet (or the outer edges of the two feet), with a margin of ≥5 cm between the pelvic projection and the sides of the support polygon, then the pelvic projection is considered stable.

[0058] In some embodiments, to make the calculation results more accurate, the motion data can be further smoothed (first-stage posture smoothing) before calculating relevant data (such as key angles) to remove high-frequency noise and jitter in the linear acceleration and angular velocity signals. This can be achieved by using a Butterworth filter for zero-phase filtering. The Butterworth filter can be of order 2 and the cutoff frequency can be 5–8 Hz; this is just an example and not a limitation.

[0059] In some embodiments, to further improve the accuracy of the calculation results, the above-mentioned pose smoothing process can be performed again after pose smoothing (second pose smoothing process).

[0060] In some embodiments, after the second pose smoothing process, median filtering can be further applied to 3–5 frames of motion data to further improve the accuracy of the calculation results.

[0061] In some embodiments, basic risk assessment indicators for each part of the body can be determined every 50ms–100ms.

[0062] In some embodiments, before the master control terminal acquires motion data and electromyography data, or before the task begins, it may send synchronization pulses to the motion capture unit and the electromyography acquisition unit to keep their time consistent with that of the master control terminal. When the human factors intelligence-based ergonomic risk monitoring system includes an event recording unit, the master control terminal may further send synchronization pulses to the event recording unit to keep the time of all devices or apparatuses in the human factors intelligence-based ergonomic risk monitoring system consistent.

[0063] In some embodiments, during task execution, the master control unit may periodically send synchronization pulses to ensure that the time of all devices or apparatuses in the human factors intelligence-based ergonomic risk monitoring system is kept as consistent as possible during task execution.

[0064] In some embodiments, before determining the basic risk assessment indicators, a light check and correction can be performed based on the aforementioned time synchronization to ensure that the motion data frames and electromyography (EMG) data frames remain on a unified time base. For example, after receiving the motion data frames and EMG data frames, the master control unit can compare the system timestamp difference between adjacent frames, the interpolated alignment frame interval, detect drift (such as an offset exceeding 1–2 ms), and fine-tune the alignment window when a slight offset is detected, so that subsequent calculations are strictly performed on a unified time axis.

[0065] Step 403: Calibrate the weights of the basic risk assessment indicators for each body part based on the electromyography (EMG) data acquired simultaneously from each body part. For example, the main control unit calibrates the weights of the basic risk assessment indicators for the corresponding body parts based on the EMG data acquired simultaneously from each body part.

[0066] For example, the master control unit can extract the temporal intensity features of each muscle group from the electromyographic data of the upper / middle trapezius, anterior deltoid, biceps brachii, triceps brachii, erector spinae, quadratus lumborum, quadriceps femoris, biceps femoris, and gastrocnemius muscles. These temporal intensity features reflect the degree of muscle activation. For example, a sliding window can be used to calculate the root mean square (RMS) of each muscle group. For example, the sliding window can be 200ms-250ms.

[0067] Afterwards, the master control unit can normalize the RMS using reference values ​​to obtain normalized electromyographic data for each muscle group.

[0068] For example, the reference value can be predetermined. This can be achieved by collecting electromyographic data of the target muscle while the operator performs a standard submaximal reference task, such as isometric abduction or flexion-extension, and using the average value as the reference value.

[0069] Then, the main control unit obtains the muscle load index of each body part based on the normalized electromyography data of each muscle group.

[0070] When a body part contains only one muscle group, or when only one electromyography (EMG) signal is collected from that body part, the muscle load index for that body part is the normalized EMG data of that muscle group. For example, if only EMG data of the upper / middle trapezius muscle is collected from the neck, then the muscle load index for the neck is the normalized EMG data of the upper / middle trapezius muscle.

[0071] When a body part has multiple muscle groups, or when electromyography (EMG) data of multiple muscle groups are collected in that body part, the muscle load index of that body part can be the maximum value among the normalized EMG data of the multiple muscle groups, or it can be the weighted sum of the normalized EMG data of the multiple muscle groups, and so on.

[0072] For example, the anterior deltoid, biceps brachii, and triceps brachii are located in the upper arm of the human body. The muscle load index of the upper arm is the maximum value of the normalized electromyography (EMG) data of the anterior deltoid, biceps brachii, and triceps brachii, or it can be the weighted sum of the normalized EMG data of the anterior deltoid, biceps brachii, and triceps brachii.

[0073] In some embodiments, before extracting the temporal intensity features of each muscle group, the electromyographic data of each muscle group can be preprocessed (first preprocessing) to make the extracted temporal intensity features more accurate.

[0074] For example, the main control unit can first perform bandpass filtering on the electromyography (EMG) data to remove DC drift and high-frequency noise, retaining only the main frequency band of muscle activity. For instance, the bandpass filtering allows the EMG data signal to pass through a frequency range of 20–450 Hz. Then, notch filtering (e.g., attenuating the EMG data signal at 50 Hz or 60 Hz) is performed to remove power frequency interference. Finally, full-wave rectification is performed to convert the signal to a positive value for subsequent amplitude feature (i.e., time-domain intensity feature) extraction.

[0075] In some embodiments, to further improve the accuracy of the temporal intensity features, a second preprocessing step may be performed after the first preprocessing. The method for the second preprocessing step is the same as that for the first preprocessing step.

[0076] After obtaining the muscle group load index of each body part, the master control terminal determines the physiological load of each body part (which can be called postural physiological load for ease of description) based on the basic risk assessment index of each body part, and determines the physiological load of each body part (which can be called electromyographic physiological load for ease of description) based on the muscle group load index of each body part.

[0077] Taking the determination of the physiological load on the upper arm as an example, the risk level (local risk level) of the upper arm can first be determined based on the basic risk assessment indicators of the upper arm. For example, the risk level of the upper arm can be obtained by multiplying the basic risk assessment indicators by a coefficient. Then, the postural physiological load matching the risk level of the upper arm can be found from a preset mapping table of risk levels and postural physiological loads; alternatively, the postural physiological load corresponding to the risk level of the upper arm can be determined through a relationship function between risk level and physiological load. This relationship function can be determined by the designer based on the actual application situation, and is not limited here.

[0078] The determination of the electromyographic (EMG) load of the upper arm can be achieved by: finding the EMG load that matches the muscle group load index of the upper arm from a pre-defined mapping table of EMG load indices and EMG physiological loads; or by determining the EMG load corresponding to the muscle group load index of the upper arm through a relationship function between a muscle group load index and physiological load. This relationship function can be determined by the designer based on the actual application, and is not limited here.

[0079] Next, the physiological load residuals for each body part are calculated. For example, the physiological load residuals for each body part can be calculated using the following formula: Δ(t) = g(E) grp (t))-f(R(t)) Wherein, g(E) grp (t) represents the electromyographic physiological load of a certain body part, f(R(t)) represents the postural physiological load of that body part, and E grpR(t) represents the muscle load index of this body part, and R(t) represents the risk level of this body part.

[0080] When Δ(t) is positive and has a large amplitude (e.g., exceeding the preset value), or when Δ(t) is positive and has a large amplitude many times within a period of time (e.g., within multiple consecutive time windows, which can be several seconds to several tens of frames), the master control terminal can determine that the actual electromyographic load is "much higher" than the posture-predicted physiological load, that is, determine that the posture-physiological load prediction for that body part is insufficient, and then increase the weight of the basic risk assessment index for that body part; conversely, the master control terminal can determine that the posture-physiological load prediction for that body part is too high, and then decrease the weight of the basic risk assessment index for that body part.

[0081] For example, when increasing or decreasing weights, the magnitude of the increase or decrease can be determined based on the size of the absolute value of the residual. For instance, a mapping table between the absolute value of the residual and the magnitude can be pre-defined, from which the magnitude matching the absolute value of the residual can be found.

[0082] For example, the initial value of the weight of the basic risk assessment index for each body part can be preset to 1. In some embodiments, the initial value of the weight of the basic risk assessment index for each body part can be the confidence level of the motion data acquisition channel for each body part. The motion data acquisition channel for each body part can be determined according to the motion capture unit configured for the body part. For example, when the motion capture unit configured for the upper arm is an IMU, there is one corresponding acquisition channel, which outputs the linear acceleration and angular velocity of the upper arm; when the motion capture unit configured for the upper arm is an optical device, there is one corresponding acquisition channel, which outputs the light spot array information; when the motion capture unit configured for the upper arm includes both an IMU and an optical device, there are two corresponding data acquisition channels.

[0083] For example, the confidence level of a particular acquisition channel can be determined based on the quality marker of the data frame transmitted from that acquisition channel to the master control terminal. The quality marker of the data frame can be determined by the master control terminal based on the quality monitoring results. Quality monitoring can be performed by the master control terminal. In some embodiments, quality monitoring can be performed before the start of the task, or it can be performed throughout the entire task process, before the master control terminal performs the aforementioned processing (such as the first posture smoothing processing, the first preprocessing) on ​​the motion data and electromyographic data.

[0084] Quality monitoring can be used to monitor whether motion data is abnormal, such as: whether acceleration / angular velocity is saturated (e.g., abnormal spikes in acceleration / angular velocity signals), whether there is magnetic interference drift (e.g., large drift in yaw angle), whether there is packet loss or jitter in acceleration / angular velocity, whether EMG electrodes are detached / whether the impedance is abnormal, whether EMG data is jittery, whether EMG data is short or missing, etc.

[0085] For example, when the master control terminal detects that a certain data frame (action data frame / EMG data frame) transmitted through a certain acquisition channel is normal, the quality flag of that data frame is set to 0; when the master control terminal detects that the data frame has packet loss or data loss, the quality flag of that data frame is set to 1; when the master control terminal detects that the data frame has saturation, the quality flag of that data frame is set to 2; and so on. Accordingly, the confidence level of the acquisition channel can be negatively correlated with the quality flag of the data frame, and the range can be 0-1. For example, when the quality flag of the data frame is 0, the confidence level of the acquisition channel can be 1; when the quality flag of the data frame is 1, the confidence level of the acquisition channel can be 0.8; when the quality flag of the data frame is 2, the confidence level of the acquisition channel can be 0.5; and so on.

[0086] In some embodiments, the master control unit can further determine, based on quality markers, whether to use strategies such as interpolation, weighting, or elimination to repair data frames. For example, when motion data or electromyography data is lost or has short gaps, the master control unit can use spline / Kalman interpolation for repair, for example, to repair short gaps of less than 1 second.

[0087] Step 404: Based on the calibrated basic risk assessment indicators and corresponding weights for each body part, determine the ergonomic risk monitoring results of the data collection object.

[0088] For example, this step may include: Step 1: The main control unit determines the overall risk level of the workers based on the basic risk assessment indicators and corresponding weights for each body part.

[0089] For example, the main control unit can first summarize the Class A comprehensive basic risk assessment indicators (hereinafter referred to as Class A indicators) and the Class B comprehensive basic risk assessment indicators (hereinafter referred to as Class B indicators), and then look up the RULA table based on the Class A and Class B indicators to obtain the overall risk level. Taking the assembly scenario as an example, Class A indicators = W 上臂 ×Basic risk assessment indicators for the upper arm + W 前臂 ×Basic risk assessment indicators for the forearm + W 手腕 × Basic risk assessment indicators for the wrist. Among them, W 上臂 As the weight of the basic risk assessment indicators for the upper arm, W 前臂 As the weight of the basic risk assessment indicators for the forearm, W 手腕 This assigns weights to the basic risk assessment indicators for the wrist. These weights are calibrated when they have been calibrated in step 403; otherwise, they are initial values. Category B Indicator = W 颈部 × Basic risk assessment indicators for the neck + W 躯干 × Basic risk assessment indicators for the torso + W 腿 ×Leg basic risk assessment indicators. Among them, W 颈部As the weight of the basic risk assessment indicators for the neck, W 躯干 W is the weight of the basic risk assessment indicators for the torso. 腿 These are the weights for the basic risk assessment indicators for the legs. When these weights are calibrated in step 403, they are calibrated weights; otherwise, they are initial values.

[0090] Then, the main control unit finds the values ​​that match the Class A and Class B indicators from the RULA table below, which represents the overall risk level.

[0091] RULA table

[0092] In the RULA table, the numbers in the first row are Class B indicators, the numbers in the first column are Class A indicators, and the bold numbers represent the risk level. For example, when the Class A indicator is 4 and the Class B indicator is 3, the current overall risk level is 4; when the Class A indicator is 5 and the Class B indicator is 6, the current overall risk level is 7; and so on.

[0093] The above RULA table is for illustrative purposes only and is not a limitation. It may be adjusted depending on the application scenario or the characteristics of the task.

[0094] Step 2: The main control unit determines the current efficiency risk based on the overall risk level.

[0095] For example, the master control unit can determine the current efficiency risk using the following formula: ERI(t) = w1 × (t)+w3×HD(t)+w4× (t)+w5×TQ(t) Among them, ERI(t) is the efficiency risk index, and the larger the index, the higher the efficiency risk. (t) represents the normalized overall risk level, H represents human height, and D(t) represents the distance function over time. (t) represents the normalized motion smoothness, T represents the corresponding index / parameter of the tool, and Q(t) is the time-varying load torque estimation function. w1, w3, w4, and w5 are respectively (t), HD(t), The weights of TQ(t) and TQ(t) are defined. w1 can be preset (e.g., set by an expert) or adjusted according to actual conditions. It can also be adaptively estimated through electromyography-posture consistency calibration (e.g., least squares + constrained learning). w3, w4, and w5 can remain at the expert setting or be adjusted independently according to the task; no limitations are imposed here. For example, w1 can be the average or weighted sum of the weights of the basic risk assessment indicators for each body part.

[0096] Compared to existing technologies that only use HD(t), The embodiments of this application further assess ergonomic risks by using TQ(t) and TQ(t) to evaluate ergonomic risks. The embodiments of this application also assess ergonomic risks by using an overall risk level calibrated by measured electromyography data, which effectively improves the accuracy of ergonomic risk monitoring.

[0097] In some embodiments, Class A and / or Class B indicators may be further integrated with any one or more of the muscle use risk assessment indicators, static retention risk assessment indicators, and load / force risk assessment indicators to further improve the accuracy of ergonomic risk monitoring.

[0098] The muscle use risk assessment index can be determined based on the muscle group load index. For example, if the upper arm muscle group load index is greater than a first threshold, or if the percentage of upper arm muscle group load indices exceeding the first threshold within a preset time period reaches a second threshold, then the muscle use risk assessment index is set to 1, and the upper arm risk assessment index equals the upper arm's baseline risk assessment index plus 1; otherwise, the muscle use risk assessment index is set to 0, and the upper arm risk assessment index remains unchanged. When the muscle use risk index is 1, the risk level of the upper arm may increase. With an increased upper arm risk level, postural physiological load may change, and the physiological load residual may also change accordingly.

[0099] Static holding risk assessment indicators and load / force risk assessment indicators can be triggered and determined by the actual operating scenario. For example, the control unit can semantically correct the scoring results by combining the contextual information of the current task (such as process stage, tool load status, operation type (grabbing / carrying / installation), hand or tool events, posture / gait changes, etc.) to ensure that the posture risk assessment is consistent with the actual operating scenario. For instance, in task segments such as "lifting," "continuous tightening," and "handling heavy parts," the control unit can automatically identify load changes, action types, and event markers in the actual scenario. When it matches the corresponding task context in the contextual information, it triggers the corresponding dynamic scoring rules (such as triggering the determination of load / force risk assessment indicators and static holding assessment indicators) to avoid different work actions being misjudged as having the same risk level at the same angle, thus making RULA's real-time score (overall risk level) closer to the actual working conditions.

[0100] Static hold evaluation metrics can be determined based on motion data. For example, the static hold evaluation metrics for the upper arm can be determined based on the dynamic data of the upper arm. For instance, if the angular velocity of the upper arm is below a threshold and the change in the torsional angle of the upper arm within a predetermined time period (e.g., 2 seconds) is less than a preset value, calculated from the motion data of the upper arm, then the static hold evaluation metrics for the upper arm are determined to be 1; otherwise, the static hold evaluation metrics for the upper arm are determined to be 0.

[0101] The load / force risk assessment index can be identified by the main control terminal based on the image. If the main control terminal identifies that the tool is turned on, the estimated torque or the lifting mass exceeds the threshold and the duration reaches the preset duration (e.g., 0.5 seconds) based on the image obtained by the event recording unit, then the load / force risk assessment index of the force-dominant part (e.g., the wrist) is determined to be 1; otherwise, the load / force risk assessment index of the force-dominant part (e.g., the wrist) is determined to be 0.

[0102] In some embodiments, when any one of the muscle use risk assessment index, static hold assessment index, and load / force risk assessment index is equal to 1, the weights of the basic risk assessment indexes do not need to be calibrated; instead, the weights of the indices with a value of 1 can be calibrated. The calibration method is the same as the weight calibration of the basic risk assessment indexes in step 403 above, except that the weights of the indices with a value of 1 among the muscle use risk assessment index, static hold assessment index, and load / force risk assessment indexes are increased or decreased. For example, the determination of the load / force risk assessment index is frequently triggered, and its corresponding characteristics (such as increased tool torque, long posture maintenance time, etc.) are highly consistent with the peak values ​​of electromyographic data, but the resulting physiological load residual Δ of the corresponding body part is not... r If (t) remains positive, the main control unit determines that the weight of the load / exertion risk assessment index is low, which is detrimental to increasing physiological load f(R). r The contribution of (t)) is insufficient, thus increasing the weight of the load / exertion risk assessment index, making its improvement in predicting physiological load more significant; when the determination of the load / exertion risk assessment index is frequently triggered, but the electromyographic physiological load does not increase significantly, and Δ r If (t) is negative within the preset time, the main control terminal determines that the weight of the load / force risk assessment index is too high, resulting in an overestimation of the postural physiological load. Therefore, the weight of the load / force risk assessment index is reduced to weaken its effect on increasing the postural physiological load.

[0103] The initial values ​​for the weights of the muscle use risk assessment index, static hold assessment index, and load / force risk assessment index can all be 1. In some embodiments, the initial values ​​for the weights of the muscle use risk assessment index can be the confidence level of the electromyography (EMG) data acquisition channel for the corresponding body part. The determination of the confidence level of the EMG data acquisition channel can be similar to the determination of the confidence level of the motion data acquisition channel, as described in the relevant descriptions in the foregoing embodiments.

[0104] In some embodiments, when any two of the muscle use risk assessment indicators, static retention assessment indicators, and load / force risk assessment indicators are equal to 1, the weight of the basic risk assessment indicator is not calibrated, and an indicator can be selected from the indicators that are equal to 1 for weight calibration.

[0105] For example, when at least two of the risk assessment indicators for muscle use, static hold, and load / force risk assessment indicators for a certain body part are equal to 1 within the same time window, the master control unit can calculate the marginal impact or explanatory power of these two indicators on postural physiological load, while keeping other conditions unchanged. The indicator with the greater marginal impact or explanatory power is then selected and its weight is calibrated. For example, assuming |Δ k (t) represents "how much residual can be eliminated if only the k-th indicator among the at least two indicators equal to 1 is adjusted", and then normalization yields the contribution ratio γ of the k-th indicator. k =|Δ k | / , where γ k Let Δ be the contribution ratio of the k-th indicator, and n be the number of indicators with at least two values ​​equal to 1. j | represents the residual that the j-th indicator among the at least two indicators equal to 1 can eliminate. γ k A larger index is likely the main source of the current residual error, so the master control unit can prioritize adjusting γ. k The weight of the largest indicator should be determined to avoid "all weights changing together," thus ensuring a stable and interpretable calibration process.

[0106] In some embodiments, the master control unit can further determine the current ergonomic risk based on muscle group load indicators. For example, the master control unit can determine the current ergonomic risk using the following formula: ERI(t) = w1 × (t)+ w2 grp(t) + w3×HD(t) + w4× (t)+w5×TQ(t) in, grp(t) is the normalized whole-body muscle load index, which is equal to the weighted sum or average of the muscle load indices of each body part. w2 is... The weights of grp(t) can be the average confidence level or weighted sum of the acquisition channels of the EMG data. For example, w2 can also be set by experts or adaptively estimated through EMG-posture consistency calibration (e.g., least squares + constraint learning).

[0107] Compared to existing technologies that only use HD(t), The embodiments of this application further assess ergonomic risks using muscle group load indicators, thereby further improving the accuracy of ergonomic risk monitoring.

[0108] Another method for monitoring work efficiency risks based on human factors intelligence provided in this application embodiment can be as follows: Figure 5As shown, it includes: Step 501: Acquire the first motion data, second motion data, and corresponding electromyographic data of the data acquisition subject during the task execution process using motion capture equipment and electromyography equipment, respectively. The first motion data is the motion data of the first part of the data acquisition subject's body, and the second motion data is the motion data of the second part of the data acquisition subject's body. The first part of the body includes the first muscle group and the second muscle group, and the second part of the body includes the third muscle group and the fourth muscle group. The electromyographic data includes the electromyographic data of the first muscle group, the second muscle group, the third muscle group, and the fourth muscle group.

[0109] The data acquisition target can be a worker in a work scenario, such as an assembly worker in an assembly scenario or a transport worker in a transport scenario. Correspondingly, the task to be performed can be assembly, transport, etc. For example, an assembly task and / or a transport task may include multiple processes. The aforementioned first motion data, second motion data, and corresponding electromyographic (EMG) data can be the motion data and EMG data acquired by the motion capture device and the EMG acquisition device respectively when the data acquisition target performs any one of these multiple processes, etc. Taking an assembly scenario as an example, before the task begins, motion capture units can be configured on the assembly worker's upper arm, forearm, wrist, neck, torso, and legs, and electrodes of an sEMG device can be configured on each muscle group in these areas. For example, an EMG acquisition channel can be configured on each of the following muscle groups: triceps brachii, biceps brachii, anterior deltoid, erector spinae, quadratus lumborum, quadriceps femoris, biceps femoris, and gastrocnemius. The first and second body parts of the data acquisition object can be any two parts from the upper arm, forearm, wrist, neck, torso, and legs. Typically, two more critical body parts are selected as the first and second parts depending on the scenario. In some embodiments, the motion data may also include motion data from more body parts of the data acquisition object, such as not only the upper arm and forearm, but also the wrist, neck, torso, and legs, etc. When the first body part is the upper arm, the first and second muscle groups can be any two muscle groups from the triceps brachii, biceps brachii, and anterior deltoid; when the first body part is the torso, the first and second muscle groups can be the erector spinae and quadratus lumborum; when the first body part is the leg, the first and second muscle groups can be any two muscle groups from the quadriceps femoris, biceps femoris, and gastrocnemius, etc. The third and fourth muscle groups are similar to the first and second muscle groups. Motion data includes linear acceleration, angular velocity, and spot array data as described in the previous embodiments. Electromyographic data includes data collected by the aforementioned sEMG device.

[0110] In this embodiment, the human factors intelligence-based ergonomic risk monitoring method can be executed by the master control terminal provided in the aforementioned embodiments. For example, after the task begins, the master control terminal acquires the first and second motion data of the data acquisition object when performing a certain procedure from the motion capture unit, and acquires the electromyographic data of the first, second, third, and fourth muscle groups from the electromyographic acquisition unit. The master control terminal can identify the corresponding body parts based on the signals transmitting the motion data. In some embodiments, the master control terminal can further name the identified body parts.

[0111] To synchronize or align the electromyographic data of the first action data, the second action data, the first muscle group, the second muscle group, the third muscle group, and the fourth muscle group, soft synchronization can be achieved through LSL.

[0112] In some embodiments, before the master control terminal acquires motion data and electromyography (EMG) data, synchronization pulses can be sent to the motion capture unit and the EMG acquisition unit to synchronize their time with the master control terminal, thereby achieving time synchronization between motion data and EMG data. When the human factors intelligence-based ergonomic risk monitoring system also includes an event recording unit, the master control terminal can further send synchronization pulses to the event recording unit to ensure that the time of the event recording unit is consistent with the time of the master control terminal.

[0113] For example, the master control unit can identify which of the following processes ("lifting," "continuous tightening," "handling heavy objects") the data acquisition object is performing based on the image information sent by the event recording unit, and match it with preset task information. The task information may include the names of all the processes within the task. In this way, the master control unit can determine from the received motion data and electromyography data which data was acquired during the first operation.

[0114] In some embodiments, the master control terminal may also receive event signals sent by the event log and determine whether the current motion data and electromyographic data belong to the motion data and electromyographic data of the first operation.

[0115] Step 502: Based on the first action data, determine the first basic risk assessment index for the first part of the body, and based on the second action data, determine the second basic risk assessment index for the second part of the body. For details, please refer to the relevant descriptions in the foregoing embodiments.

[0116] Step 503: Based on the electromyographic data of the first muscle group and the second muscle group, determine the load index of the first muscle group; based on the electromyographic data of the third and fourth muscle groups, determine the load index of the second muscle group. The method for determining the muscle group load index is described in the relevant instructions in the preceding embodiments.

[0117] For example, determining the load index of the first muscle group based on the electromyographic data of the first muscle group and the electromyographic data of the second muscle group may include: Extract the first temporal intensity feature from the electromyography data of the first muscle group, and extract the second temporal intensity feature from the electromyography data of the second muscle group; The intensity features in the first and second time domains were standardized to obtain normalized electromyographic data for the first and second muscle groups. The load index of the first muscle group is determined based on the normalized electromyography (EMG) data of the first and second muscle groups. For further details on determining the load index of the first muscle group based on the normalized EMG data of the first and second muscle groups, please refer to the relevant descriptions in the foregoing embodiments.

[0118] The determination of the load index for the second muscle group can be similar to that for the first muscle group.

[0119] Steps 502 and 503 can be performed in parallel.

[0120] Step 504: Based on the first muscle group load index, calibrate the first weight, which is the weight of the first basic risk assessment index; based on the second muscle group load index, calibrate the second weight, which is the weight of the second basic risk assessment index.

[0121] For example, calibrating the first weight based on the first muscle group load index may include: Based on the first basic risk assessment indicator, the first local risk level is determined, which is the risk level of the first part of the body. Based on the first muscle group load index and the first local risk level, the first load residual is determined, which is the load residual of the first part of the body; When the value of the first load residual is positive, the first weight is increased; when the value of the first load residual is negative, the first weight is decreased.

[0122] For example, the first load residual can be determined using the following formula: Δ1(t)=g(Egrp1(t))-f(R1(t)) Where Δ1(t) is the first load residual, Egrp1(t) is the first muscle group load index, R1(t) is the first local risk level, g(Egrp1(t)) is the physiological load determined based on Egrp1(t), and f(R1(t)) is the physiological load determined based on R1(t). When Δ1(t) > 0, the first weight is increased; when Δ1(t) < 0, the first weight is decreased. The first weight is calibrated based on the first muscle group load index, as can be found in the relevant descriptions in the foregoing embodiments.

[0123] In some embodiments, before calibrating the first weight based on the first muscle group load index, the following may also be included: The risk assessment index for the use of the first muscle is determined based on the load index of the first muscle group. The risk assessment index for the use of the first muscle is the risk assessment index for the use of the first muscle in the body. Based on the first motion data, a first static hold risk assessment index and a first load / force risk assessment index are determined. The first static hold risk assessment index is the static hold risk assessment index of the first part of the body, and the first load / force risk assessment index is the load / force risk assessment index of the first part of the body. Please refer to the relevant descriptions in the foregoing embodiments. Accordingly, calibrating the first weight based on the first muscle group load index may include: If the first muscle use risk assessment index, the first static hold risk assessment index, and the first load / force risk assessment index all fail to meet the conditions, the first weight is calibrated based on the first muscle group load index, as described in the relevant descriptions in the foregoing embodiments.

[0124] In some embodiments, if any one of the first muscle use risk assessment index, the first static retention risk assessment index, and the first load / force risk assessment index meets the condition (e.g., any index is 1), the first weight remains unchanged, as can be seen from the relevant description in the foregoing embodiments; Accordingly, before determining the overall risk level of the data collection object during task execution based on the first basic risk assessment indicator, the second basic risk assessment indicator, and the calibrated first and second weights, the following may also be included: The third weight is calibrated. The third weight is the weight of the third indicator. The third indicator is any one of the first muscle use risk assessment indicator, the first static retention risk assessment indicator, and the first load / force risk assessment indicator that meets the conditions. Please refer to the relevant description in the foregoing embodiments. Accordingly, based on the first basic risk assessment indicator, the second basic risk assessment indicator, and the calibrated first and second weights, the overall risk level of the data collection object during the task execution process is determined, which may include: Based on the first basic risk assessment indicator, the second basic risk assessment indicator, the first weight, the calibrated second weight, the third indicator, and the calibrated third weight, the overall risk level of the data collection object during the task execution process is determined, as can be found in the relevant descriptions in the aforementioned embodiments.

[0125] In some embodiments, calibrating the third weight may include: Based on the first basic risk assessment indicator and the third indicator, the first local risk level is determined. The first local risk level is the risk level of the first part of the body, as can be seen from the relevant description in the foregoing embodiments. Based on the first muscle group load index and the first local risk level, the first load residual is determined, which is the load residual of the first part of the body; When the value of the first load residual is positive, the third weight is increased; when the value of the first load residual is negative, the third weight is decreased.

[0126] For example, the first load residual can be determined using the following formula: Δ1(t)=g(Egrp1(t))-f(R1(t)) Where Δ1(t) is the first load residual, Egrp1(t) is the first muscle group load index, g(Egrp1(t)) is the physiological load corresponding to Egrp1(t), R1(t) is the first local risk level, and f(R1(t)) is the physiological load corresponding to R1(t). When Δ1(t) > 0, the third weight is increased; when Δ1(t) < 0, the third weight is decreased. Please refer to the relevant description in the aforementioned embodiments.

[0127] In some embodiments, if any two of the first muscle use risk assessment index, the first static retention risk assessment index, and the first load / force risk assessment index meet the conditions (e.g., both are 1), the first weight remains unchanged, as can be seen from the relevant description in the foregoing embodiments; Accordingly, before determining the overall risk level of the data collection object during task execution based on the first basic risk assessment indicator, the second basic risk assessment indicator, and the calibrated first and second weights, the following may also be included: Determine the first load impact index and the second load impact index. The first load impact index is the impact index of the fourth index on the load, and the second load impact index is the impact index of the fifth index on the load. The fourth index and the fifth index can be any two indices. See the relevant description in the foregoing embodiments. When the first load impact index is greater than the second load impact index, the fourth weight is calibrated. The fourth weight is the weight of the fourth index, as can be seen in the relevant description in the aforementioned embodiments. Based on the first basic risk assessment indicator, the second basic risk assessment indicator, and the calibrated first and second weights, the overall risk level of the data collection object during task execution is determined, including: Based on the first basic risk assessment indicator, the second basic risk assessment indicator, the first weight, the calibrated second weight, the fourth indicator, the fifth indicator, the fifth weight, and the calibrated fourth weight, the overall risk level of the data collection object during the task execution process is determined. The fifth weight is the weight of the fifth indicator, as can be seen in the relevant description in the aforementioned embodiments.

[0128] In some embodiments, determining the first load impact index and the second load impact index may include: Based on the load residuals before and after the change of the fourth indicator, the first load impact indicator is determined. For example, the first load impact indicator can be determined using the following formula: Δ40= g(Egrp1(t))-f(R10(t)) Δ41= g(Egrp1(t))-f(R11(t)) |Δ4|=|Δ41-Δ40| Wherein, Egrp1(t) is the first muscle group load index, g(Egrp1(t)) is the physiological load corresponding to Egrp1(t), R10(t) is the first local risk level determined based on the first basic risk assessment index and the fourth index, the first local risk level is the risk level of the first part of the body, R11(t) is the first local risk level determined based on the first basic risk assessment index and the changed fourth index, f(R10(t)) is the physiological load determined based on R10(t), f(R11(t)) is the physiological load determined based on R11(t), Δ40 is the load residual corresponding to the fourth index before the change, Δ41 is the load residual corresponding to the fourth index after the change, and |Δ4| is the first load influence index, which can also be understood as the residual that the fourth index can eliminate; Based on the load residuals before and after the change of the fifth indicator, the second load impact indicator is determined. For example, the second load impact indicator can be determined using the following formula: Δ50= g(Egrp1(t))-f(R10(t)) Δ51= g(Egrp1(t))-f(R11(t)) |Δ5|=|Δ51-Δ50| Wherein, Egrp1(t) is the first muscle group load index, g(Egrp1(t)) is the physiological load corresponding to Egrp1(t), R10(t) is the first local risk level determined based on the first basic risk assessment index and the fifth index, the first local risk level is the risk level of the first part of the body, R11(t) is the first local risk level determined based on the first basic risk assessment index and the modified fifth index, f(R10(t)) is the physiological load determined based on R10(t), f(R11(t)) is the physiological load determined based on R11(t), Δ50 is the load residual corresponding to the fifth index before the change, Δ51 is the load residual corresponding to the fifth index after the change, and |Δ5| is the second load influence index, which can also be understood as the residual that the fifth index can eliminate.

[0129] In some embodiments, the first load impact index and the second load impact index may be further normalized, such as... y4 = |Δ4| / (|Δ4| + |Δ5|) y5 = |Δ5| / (|Δ4| + |Δ5|) Wherein, y4 is the normalized first load impact index, and y5 is the normalized second load impact index.

[0130] The determination of the first load impact index and the second load impact index can also be found in the relevant descriptions in the foregoing embodiments.

[0131] Step 505: Based on the first basic risk assessment indicator, the second basic risk assessment indicator, and the calibrated first and second weights, determine the overall risk level of the data collection object during the task execution process. See the relevant description in the foregoing embodiments.

[0132] Step 506: Based on the overall risk level, determine the efficiency risk of the data collection object during the task execution process, as can be found in the relevant descriptions in the aforementioned embodiments.

[0133] In some embodiments, the conditions for the first muscle group to meet the risk assessment index may include: the first muscle group load index is a high load index, or the proportion of the first muscle group load index that is a high load index within a preset time period reaches a first threshold, and the high load index is the first muscle group load index that exceeds a second threshold. Please refer to the relevant descriptions in the foregoing embodiments.

[0134] In some embodiments, the human factors intelligence-based ergonomic risk monitoring method may further include: displaying the aforementioned first local risk level, or displaying the first local risk level and the second local risk level, or displaying more local risk levels, such as simultaneously displaying the risk levels of the upper arm, forearm, wrist, neck, torso, and legs. The display can be in the form of text information or images, such as... Figure 6 As shown, the darkest color on the upper arm indicates the highest risk level, followed by the torso, then the neck, and so on.

[0135] The second local risk level refers to the risk level of the second part of the body, and the method for determining it is similar to that of the first local risk level. The determination of the risk level of the remaining parts of the body is also similar to that of the first local risk level.

[0136] In some embodiments, before determining the efficiency risk of the data collection object during task execution based on the overall risk level, the following may also be included: Based on the confidence level of the data acquisition channels, a sixth weight is determined. This sixth weight represents the overall risk level. The data acquisition channels include channels for acquiring motion data and channels for acquiring electromyographic data, as described in the previous embodiments. For example, the sixth weight can be derived from the aforementioned ERI(t) formula. The weight w1 of (t); Accordingly, based on the overall risk level, the efficiency risks of the data collection objects during task execution can be determined, which may include: Based on the overall risk level and the sixth weight, the efficiency risk is determined, as detailed in the relevant descriptions in the aforementioned embodiments.

[0137] In some embodiments, determining the work efficiency risk of the data collection object during task execution based on the overall risk level may include: determining the work efficiency risk based on the first muscle group load index, the second muscle group load index, and the overall risk level, as detailed in the relevant descriptions in the foregoing embodiments.

[0138] The aforementioned determination of efficiency risk refers to the current efficiency risk. In some embodiments, after determining the efficiency risk, the efficiency risks of different time periods can be further accumulated for task / process-level ranking. For example, this can be accumulated using the formula: CPL = ∫ ERI(t)dt. The accumulated efficiency risk CPL can be accumulated in a tiered manner, such as a short-term accumulation window of 60 seconds to reflect the accumulation trend of instantaneous load; process-level accumulation covers the entire process duration; and task-level accumulation can cover the entire task or shift to assess medium- and long-term efficiency risks.

[0139] In this embodiment, task or process-level ranking is based on ergonomic risk / competitive liability (ERI / CPL). The core idea is to use quantitative indicators to reflect the degree of human workload and injury risk, thereby identifying the "most dangerous / highest priority intervention" tasks and processes. The higher the ergonomic risk of a task / process, the higher its priority, indicating that the task has a greater potential to cause harm to the human body and should be prioritized for design optimization or job reassignment.

[0140] In some embodiments, after the first preprocessing and before the second preprocessing, the process may further include: extracting temporal features, such as RMS, from the electromyography data for judging and labeling the quality of the electromyography / motion data.

[0141] In some embodiments, after the first preprocessing and before the second preprocessing, the process may further include: extracting frequency domain features from the electromyography (EMG) data, such as median frequency (MDF), mean frequency (MNF), and average absolute value (ARV), to supplement the fatigue trend, muscle activation pattern, and noise recognition capabilities that time-domain RMS cannot reflect. For example, the main control unit can monitor the continuous decreasing trend of MDF / MNF (e.g., a decrease exceeding the initial reference ratio by 10%–20%) within a 5–10 second sliding window, and combine this with changes in RMS or ARV to identify whether muscle fatigue is increasing. Fatigue alerts can be presented through timeline color markings, real-time graphs, or text alarms to help assess the potential risks caused by continuous high load.

[0142] In some embodiments, frequency domain features extracted from electromyography (EMG) data can also be used to enhance quality monitoring, such as identifying motion artifacts and low-frequency noise in EMG data.

[0143] In some embodiments, after the first posture smoothing process and before the second posture smoothing process, the process may further include: identifying body parts (also known as RULA sub-entries) based on motion data signals, such as the upper arm, forearm, wrist, neck, torso, and legs, and naming the identified body parts based on human anatomy to obtain the worker's skeletal system.

[0144] In some embodiments, after the first pose smoothing process and before the second pose smoothing process, the process may further include: adopting an internationally recognized right-handed coordinate system (Z-up) unified skeleton system and performing skeleton normalization based on the initial pose of the motion, so as to uniformly map pose data from different data sources to a standard skeleton template (such as Biovision Hierarchy or MakeHuman skeleton structure) to ensure consistency with the virtual human platform (such as Jack).

[0145] In some embodiments, the central control unit can also receive user-inputted information such as height, weight, personal protective equipment (PPE) constraints, and tool parameters. Based on this information and the acquired motion data, it calculates personalized parameters such as limb length, joint zero position, and accessibility, and then generates a personalized virtual human based on these personalized parameters. The user-inputted information provides physiological and task constraint information, while the motion data provides body geometry and kinematic information. The fusion of these two can be used to construct personalized virtual humans and optimization models. Compared to existing virtual human simulation platforms that rely solely on geometric parameters for personalization, this application embodiment uses geometric parameters calculated from motion data and user-inputted parameters, along with posture risk and electromyography calibration, to perform joint individualized modeling. This ensures that the virtual human conforms to both structural proportions and real muscle usage patterns. Simultaneously, it avoids the high costs of traditional inverse dynamics and manual parameter tuning, making virtual human modeling more accurate, lightweight, and cost-effective.

[0146] In some embodiments, when the two are integrated, the central control unit can also construct a parametric workstation model based on the platform height H, fixture rotation angle θ, workpiece position (x,y,z), tool suspension height h, and reachable area radius r. The parametric workstation model can be constructed using existing methods, but this application embodiment deeply couples it with personalized virtual humans, posture-electromyography consistency calibration, real-time ergonomic risk calculation, and multi-objective optimization. This allows the parametric model to be used not only for simulation but also for automatically deriving the optimal workstation adjustment scheme, thus forming a new application mode.

[0147] In some embodiments, after the master control terminal obtains the measured posture based on the motion data, it can also drive a personalized virtual human based on the measured posture to calculate joint torque, accessibility, field of view occlusion, collision, etc. in the simulation platform (such as Jack).

[0148] In some embodiments, the main control terminal can also determine the ERI / CPL of each joint or body part based on the risk level (local risk level) of the worker's body parts, and map it onto the virtual human model with corresponding color or color depth to obtain a human body heat map. Figure 6 As shown, the highlighted areas are high-risk areas. For example, if the shoulder / wrist is too red (the darkest color), it indicates a high risk of posture.

[0149] In some embodiments, the main control terminal can also draw the ERI(t) curve to find the time point when the peak occurs, and with the help of virtual animation, it can trace back what specific posture / operation led to the risk increase, so as to use it for "review + cause analysis + optimization point annotation".

[0150] Furthermore, human body heatmaps and ERI(t) curves can be used for spatial-temporal visualization diagnostics, enabling users to more intuitively identify problems and thus make optimizations. For example, a user might discover a sharp increase in shoulder heat at the 12th second using a human body heatmap, then check the animation and find that the tool position is too high. Subsequently, the layout can be optimized by lowering the tool position.

[0151] In some embodiments, after determining the efficiency risk of the data collection object in the process of performing the task based on the overall risk level, the method may further include: performing reverse optimization based on the efficiency risk to generate improvement suggestions.

[0152] For example, reverse optimization can be performed by an optimizer in the master control unit. The optimization objective can be to minimize the peak value of CPL or ERI. t While calculating ERI(t), constraints are imposed on cycle time, human accessibility, tooling interference, and load torque. The optimization algorithm can be a heuristic search for discrete / continuous mixed variables (such as NSGA-II / Bayesian optimization / simulated annealing), outputting a Pareto solution set. During optimization, key variables such as human height H, joint posture angle θ, tool layout position (x,y,z), stance parameter zh, table height, and tool rotation angle are first input to the optimizer. These variables constitute a high-dimensional search space used to simulate possible human-machine configuration schemes. The optimizer can evaluate multiple objective functions in each round through iterative search, including the cumulative exposure risk index (CPL), instantaneous ergonomic risk peak ERI(t), human accessibility (determining whether the operator can do it (spatial feasibility)), posture deviation degree (determining the risk when the operator does it (posture rationality)), load torque, cycle time, and tooling interference. The performance of each solution across multiple objectives is non-dominated and sorted, ultimately outputting a Pareto solution set, which is a set of "optimal compromise solutions" that cannot be further optimized on any one objective without sacrificing other objectives. Heuristic search is responsible for "global exploration"; iterative algorithms are responsible for "local refinement"; and heuristic multi-objective optimization is responsible for "multi-objective trade-offs and Pareto selection".

[0153] Pareto solutions represent alternative configurations with different optimization preferences. For example, some solutions are better at reducing CPL but have slightly higher posture deviations, while others ignore some risk controls while maintaining good human accessibility. The system can combine constraints in the production scenario (such as manufacturing feasibility, tooling layout limitations, material costs, downtime, etc.) to perform weighted filtering and ranking of the solutions in the Pareto solution set, and automatically recommend the Top-N optimal configurations for implementation. For example, the optimizer may output a specific optimization suggestion: "Table height 800mm, operating angle rotated 12° clockwise, tool attachment point moved forward 100mm", and further compare the CPL reduction and ERI peak change of this solution in the simulation environment with the initial working conditions to help verify its optimization value and implementation potential.

[0154] The selection and ranking of solutions in the Pareto solution set can be combined with a score based on production feasibility costs (modification difficulty, downtime, material costs), automatically recommending Top-N solutions. Personalized suggestions at the top of the Top-N solutions include "increase table height by 80mm, rotate fixture clockwise by 12°, move tool suspension point forward by 100mm." Furthermore, comparisons of indicators before and after optimization can be provided (e.g., ...). Figure 7 (as shown) and the estimated decrease (as shown) Figure 8 (As shown).

[0155] like Figure 7 As shown, curve 71 shows the change of the human ergonomic risk index over time before optimization, and curve 72 shows the change of the simulated human ergonomic risk index over time after optimization. It can be seen that curve 72 decreases overall and is more stable, indicating that the risk peak is reduced and the posture is more stable through the adjustment of work position / tool.

[0156] like Figure 8 As shown, column 811 represents the maximum instantaneous risk before optimization, 821 represents the CPL before optimization, and 831 represents the joint deviation before optimization. Column 812 represents the simulated maximum instantaneous risk after optimization, 822 represents the simulated CPL after optimization, and 832 represents the simulated joint deviation after optimization. It can be seen that the optimized scheme not only reduces the instantaneous dangerous posture but also alleviates the long-term cumulative load, while improving movement coordination.

[0157] Furthermore, human body surface heat maps can visually demonstrate whether the risk of different joints or body areas has decreased. Assuming the human body surface heat map before optimization is as follows: Figure 6 As shown, the shoulder, neck, and wrist joint areas are darker, indicating concentrated load and high risk in these areas. After optimization, the color of the human body surface heat map is significantly reduced, demonstrating that the optimized adjustment effectively alleviated the high-risk areas and improved overall postural efficiency.

[0158] When the optimization scheme is deemed effective through simulation or modeling, visualization / reports can be exported, and structured data such as optimization parameters, KPI indicators, risk reduction, and report indexes can be written back to the MES / IE system via API.

[0159] An example of the data structure involved in this application is as follows: Subject{id, anthropometry{height, arm, forearm, hand, ...},strengthLevel} MoCapFrame{t, joints{neck, trunk, shoulder_L, elbow_L, wrist_L, ...},quat / angle} EMGFrame{t, channels{UT_L, UT_R, ES_L, ES_R, ...}, rms, mdf, quality} Event{t, type{task_start, tool_on, lift, ...}, meta{loadNm, toolId}} RULAFrame{t, items{upperArm, lowerArm, wrist, neck, trunk, legs},score, adjusted} RiskFrame{t, R_adj, E_grp, HD, J, TQ, ERI} Workcell{H, theta, x, y, z, h, r, ...} OptSolution{id, vars{H,theta,...}, KPI{peakERI, CPL, cycle, cost}} The write-back of structured data can exist alongside visualization / report export, and it is a data interaction process in which the central control terminal automatically pushes the analysis results to the MES / IE system through an interface.

[0160] The human factors intelligence-based ergonomic risk monitoring method provided in this application is applied to the following scenario: During the assembly of a machine tool's "fixed shell" process, before optimization, the average RULA value was 5.6, with a peak value of 6; the peak value of UT electromyography was approximately 0.42 MVC, and the peak value of ERI was 0.78. After optimization, the following adjustment scheme was proposed: table height +80mm, fixture rotation 12°, and tool suspension lowered by 90mm. After adjustments were made according to the scheme, the human factors intelligence-based ergonomic risk monitoring method provided in this application was used again for monitoring. The results were: shoulder abduction angle ↓14°, neck flexion ↓8°, UT peak EMG prediction ↓27%, ERI peak ↓32%, and CPL ↓29%.

[0161] When the human factors intelligence-based ergonomic risk monitoring method provided in this application embodiment is applied to the scenario of organizing a route (high cognitive / medium physical exertion), after optimization by adding micro-rests / segmented tabletops and optimizing viewing distance, the human factors intelligence-based ergonomic risk monitoring method provided in this application embodiment is used for monitoring. The results are: the neck RULA sub-item changes from 3 to 2, and CPL decreases by 18%.

[0162] When the human factors intelligence-based ergonomic risk monitoring method provided in this application embodiment is applied to the scenario of installing the Z-axis lower guard plate (high physical strength / medium cognitive strength), after optimization: the foot pedal height is increased to be adjustable and the workpiece is moved forward by 60mm, the lower back bending angle is reduced, and then the human factors intelligence-based ergonomic risk monitoring method provided in this application embodiment is used for monitoring. The result is: the peak value of erector spinae muscle EMG is ↓23%.

[0163] The ergonomic risk monitoring method, equipment, and system provided in this application improve the consistency between risk assessment and real muscle load through electromyography-posture consistency calibration and personalized virtual humans. Furthermore, visualization is enhanced, presenting risk hotspots and causes in a heatmap / timeline format within the virtual human body and the 3D workstation environment, supporting source tracing. Further, through reverse optimization, parameterized executable solutions are directly output, providing implementable suggestions and verifying the improvement effects in simulation, shortening the path from "problem discovery → implementation of changes." Moreover, it supports rapid on-site assessment (simultaneous data collection and calculation) for in-cycle decision-making or post-shift review, exhibiting high real-time performance. Furthermore, the multi-source quality monitoring, mismatch repair, and interface-based architecture facilitate integration with existing MES / PLM / IE toolchains, adapting to different workstations and personnel. In addition, the middle layer achieves module decoupling and cross-engine calculation through algorithm and simulation interfaces, while the downstream layer achieves result write-back and business linkage through enterprise system interfaces, demonstrating high robustness and scalability. Among them, mismatch repair refers to the system automatically detecting and correcting the deviations between the virtual model and the real workstation and human-machine parameters after generating optimization suggestions, so that the virtual human is consistent with reality, and prepares for subsequent transformation based on the optimization suggestions.

[0164] The edge computing device provided in this application embodiment may include a processor, a memory, and a computer program / instructions. There may be one or more processors, coupled to the memory, and the computer program is stored in the memory. When the processor calls and executes the computer program in the memory, the edge computing device can perform the ergonomic risk monitoring method provided in any of the above embodiments.

[0165] like Figure 9 As shown, the edge computing device 90 includes a processor 91, a memory 92, and a computer program 93. The computer program 93 is stored in the memory 92, and the processor 91 is coupled to the memory 92. When the processor 91 calls and executes the computer program 93 in the memory 92, the edge computing device 90 can execute the ergonomic risk monitoring method provided in any of the above embodiments.

[0166] The computer-readable storage medium provided in this application embodiment may be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc. This computer-readable storage medium may store a computer program, and when the computer program is run on a computer, the computer can execute the methods provided in any of the above embodiments.

[0167] It should be understood that the division of modules in the above embodiments is merely a logical functional division. In actual implementation, they can be fully or partially integrated into a single physical entity, or they can be physically separated. Furthermore, these modules can be implemented entirely in hardware; alternatively, some modules can be implemented in software through processing element calls, while others are implemented in hardware. For example, the processor can be a separate processing element or integrated into a chip within the electronic device. The implementation of other modules is similar. Moreover, these modules can be fully or partially integrated together, or implemented independently. During implementation, each of the above modules can be completed through integrated logic circuits in the hardware of the processor element or through software instructions.

[0168] In this application embodiment, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent the existence of A alone, A and B simultaneously, or B alone. A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of singular or plural items. For example, at least one of a, b, and c can represent: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, and c can be single or multiple.

[0169] The above description is merely a specific embodiment of this application. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the protection scope of this application. The protection scope of this application should be determined by the protection scope of the claims.

Claims

1. A method for monitoring work efficiency risks based on human factors intelligence, characterized in that, include: The motion capture device and electromyography device are used to acquire motion data and electromyography data of at least two body parts of the data acquisition object during the performance of the task. Based on the motion data, basic risk assessment indicators for each body part of the data collection object are determined; The weights of the basic risk assessment indicators for each body part are calibrated based on the electromyography data of each body part acquired at the same time. Based on the calibrated basic risk assessment indicators and corresponding weights for each body part, the ergonomic risk monitoring results of the data collection object are determined.

2. The method according to claim 1, characterized in that, The step of acquiring motion data and electromyographic data of at least two body parts of the data acquisition object during the execution of a task using motion capture equipment and electromyography (EMG) equipment includes: acquiring first motion data, second motion data, and corresponding EMG data of muscle groups of the body parts of the data acquisition object during the execution of a task using motion capture equipment and EMG equipment, respectively. The first motion data is the motion data of a first body part of the data acquisition object, and the second motion data is the motion data of a second body part of the data acquisition object. The first body part includes a first muscle group and a second muscle group, and the second body part includes a third muscle group and a fourth muscle group. The EMG data of the muscle groups includes the EMG data of the first muscle group, the EMG data of the second muscle group, the EMG data of the third muscle group, and the EMG data of the fourth muscle group. The step of determining the basic risk assessment indicators for each body part of the data collection object based on the motion data includes: determining the first basic risk assessment indicator for the first body part based on the first motion data, and determining the second basic risk assessment indicator for the second body part based on the second motion data. The step of calibrating the weights of the basic risk assessment indicators for each body part based on the electromyographic data of each body part acquired simultaneously includes: determining a first muscle group load indicator based on the electromyographic data of the first muscle group and the second muscle group; determining a second muscle group load indicator based on the electromyographic data of the third muscle group and the fourth muscle group; calibrating a first weight based on the first muscle group load indicator, wherein the first weight is the weight of the first basic risk assessment indicator; and calibrating a second weight based on the second muscle group load indicator, wherein the second weight is the weight of the second basic risk assessment indicator. The determination of the work efficiency risk assessment result of the data collection object based on the calibrated basic risk assessment indicators and corresponding weights for each body part includes: determining the overall risk level of the data collection object in the process of performing the task based on the first basic risk assessment indicator, the second basic risk assessment indicator, and the calibrated first weight and second weight; and determining the work efficiency risk monitoring result of the data collection object in the process of performing the task based on the overall risk level.

3. The method according to claim 2, characterized in that, The calibration of the first weight based on the first muscle group load index includes: Based on the first basic risk assessment index, a first local risk level is determined, and the first local risk level is the risk level of the first part of the body. Based on the first muscle group load index and the first local risk level, a first load residual is determined, wherein the first load residual is the load residual of the first part of the body. When the value of the first load residual is positive, the first weight is increased; when the value of the first load residual is negative, the first weight is decreased.

4. The method according to claim 2, characterized in that, Before calibrating the first weight based on the first muscle group load index, the method further includes: The first muscle use risk assessment index is determined based on the first muscle group load index. The first muscle use risk assessment index is the muscle use risk assessment index of the first part of the body. Based on the first action data, a first static hold risk assessment index and a first load / force risk assessment index are determined. The first static hold risk assessment index is the static hold risk assessment index of the first part of the body, and the first load / force risk assessment index is the load / force risk assessment index of the first part of the body. The calibration of the first weight based on the first muscle group load index includes: If the first muscle use risk assessment index, the first static holding risk assessment index, and the first load / force risk assessment index all fail to meet the conditions, the first weight is calibrated based on the first muscle group load index.

5. The method according to claim 4, characterized in that, If any one of the first muscle use risk assessment index, the first static retention risk assessment index, and the first load / force risk assessment index meets the conditions, the first weight remains unchanged; Before determining the overall risk level of the data collection object during the execution of the task based on the first basic risk assessment indicator, the second basic risk assessment indicator, and the calibrated first and second weights, the method further includes: The third weight is calibrated, which is the weight of the third indicator. The third indicator is any one of the following indicators that meets the conditions: the first muscle use risk assessment indicator, the first static retention risk assessment indicator, and the first load / force risk assessment indicator. The step of determining the overall risk level of the data collection object during the execution of the task based on the first basic risk assessment indicator, the second basic risk assessment indicator, and the calibrated first and second weights includes: Based on the first basic risk assessment indicator, the second basic risk assessment indicator, the first weight, the calibrated second weight, the third indicator, and the calibrated third weight, the overall risk level of the data collection object during the execution of the task is determined.

6. The method according to claim 5, characterized in that, The calibration of the third weight includes: Based on the first basic risk assessment indicator and the third indicator, a first local risk level is determined, and the first local risk level is the risk level of the first part of the body. Based on the first muscle group load index and the first local risk level, a first load residual is determined, wherein the first load residual is the load residual of the first part of the body. When the value of the first load residual is positive, the third weight is increased; when the value of the first load residual is negative, the third weight is decreased.

7. The method according to claim 4, characterized in that, If any two of the first muscle use risk assessment index, the first static holding risk assessment index, and the first load / force risk assessment index meet the conditions, the first weight remains unchanged. Before determining the overall risk level of the data collection object during the execution of the task based on the first basic risk assessment indicator, the second basic risk assessment indicator, and the calibrated first and second weights, the method further includes: Determine the first load impact index and the second load impact index, wherein the first load impact index is the fourth index's impact index on the load, and the second load impact index is the fifth index's impact index on the load, wherein the fourth index and the fifth index are any two of the aforementioned indices; When the first load impact index is greater than the second load impact index, the fourth weight is calibrated, whereby the fourth weight is the weight of the fourth index. The step of determining the overall risk level of the data collection object during the execution of the task based on the first basic risk assessment indicator, the second basic risk assessment indicator, and the calibrated first and second weights includes: The overall risk level of the data collection object in the process of performing the task is determined based on the first basic risk assessment indicator, the second basic risk assessment indicator, the first weight, the calibrated second weight, the fourth indicator, the fifth indicator, the fifth weight, and the calibrated fourth weight, wherein the fifth weight is the weight of the fifth indicator.

8. The method according to claim 7, characterized in that, The determination of the first load impact index and the second load impact index includes: Based on the load residual before the fourth indicator was changed and the load residual after the fourth indicator was changed, the first load impact indicator was determined. The second load impact index is determined based on the load residual before the change of the fifth index and the load residual after the change of the fifth index.

9. The method according to claim 5 or 7, characterized in that, The first muscle group load indicator meets the following conditions: the first muscle group load indicator is a high load indicator, or the proportion of the first muscle group load indicator as a high load indicator within a preset time period reaches a first threshold, wherein the high load indicator is a first muscle group load indicator that exceeds a second threshold.

10. The method according to claim 2, characterized in that, Before determining the ergonomic risk monitoring results of the data collection object during the execution of the task based on the overall risk level, the method further includes: Based on the confidence level of the data acquisition channels, a sixth weight is determined, which is the weight of the overall risk level. The data acquisition channels include the acquisition channels for motion data and the acquisition channels for electromyographic data. The determination of the work efficiency risk monitoring results of the data collection object during the execution of the task based on the overall risk level includes: The ergonomic risk monitoring results are determined based on the overall risk level and the sixth weight; or, The determination of efficiency risks for data collection objects during the execution of the task based on the overall risk level includes: The ergonomic risk monitoring results are determined based on the first muscle group load index, the second muscle group load index, and the overall risk level; or, After determining the efficiency risk monitoring results of the data collection object during the execution of the task based on the overall risk level, the method further includes: Based on the results of the ergonomic risk monitoring, reverse optimization is performed to generate improvement suggestions.

11. The method according to any one of claims 2-8, characterized in that, The determination of the load index for the first muscle group based on the electromyographic data of the first muscle group and the electromyographic data of the second muscle group includes: Extract a first temporal intensity feature from the electromyography data of the first muscle group, and extract a second temporal intensity feature from the electromyography data of the second muscle group; The first time-domain intensity features and the second time-domain intensity features are standardized respectively to obtain normalized electromyographic data of the first muscle group and the second muscle group. Based on the normalized electromyographic data of the first muscle group and the second muscle group, the load index of the first muscle group is determined.

12. An edge computing device, characterized in that, include: One or more processors and a memory, the one or more processors being coupled to the memory for storing a computer program, wherein when the one or more processors execute the computer program, the edge computing device performs the method as described in any one of claims 1-11.

13. A work efficiency risk monitoring system based on human factors intelligence, characterized in that, The device includes a main control terminal, a motion capture unit, and an electromyography (EMG) acquisition unit. The main control terminal is connected to the motion capture unit and the EMG acquisition unit, respectively. The motion capture unit is used to acquire motion data, and the EMG acquisition unit is used to acquire EMG data. The main control terminal is used to execute the method as described in any one of claims 1-11.

14. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when run on a computer, enables the computer to perform the method described in any one of claims 1-11.

Images