Human ergonomics risk analysis method and system for workers based on digital twinning

Abstract

The invention belongs to the related technical field of digital twinning, and discloses a worker ergonomics risk analysis method and system based on digital twinning. The method comprises the following steps of: converting a spatio-temporal trajectory of a worker into a nested network of a region transfer network and an attitude transformation network; respectively calculating the number of nodes, the average node intensity and the transfer degree of the regional transfer network and the attitude transformation network; and respectively calculating the human ergonomics risk related to a single action and the human ergonomics risk related to a mechanical cycle action by using the node number, the average node intensity and the transfer degree of the regional transfer network and the attitude transformation network so as to realize human ergonomics risk analysis. According to the invention, the influence of the worker attitude, the environmental factor and the mechanical cycle operation is comprehensively considered, a comprehensive worker risk identification model is constructed, and accurate risk monitoring and management are realized.

Description

Technical Field

[0001] The present invention belongs to the technical field related to digital twins, and more specifically, relates to a worker ergonomic risk analysis method and system based on digital twins. Background Art

[0002] As the construction environment becomes more complex, the interaction between workers' posture and the environment during work has attracted more and more attention. By using digital twin technology and workers' somatosensory data, the engineering field is exploring how to monitor workers' posture and risk factors in real time to reduce ergonomic risks caused by poor posture or environmental factors. In the current construction process, the ergonomic risks faced by workers mainly come from the interaction of poor posture, repetitive mechanical work and environmental factors, but most of the existing monitoring technologies only focus on a single factor and it is difficult to fully identify these complex risks. On the one hand, existing research mainly relies on machine learning models to identify workers' postures, but the data collection cost is high and ignores the environmental impact of workers during construction, which leads to the omission of some potential musculoskeletal disease risks. On the other hand, current research rarely involves the identification of repetitive and monotonous mechanical cycle operations, and long-term mechanical cycle operations may lead to musculoskeletal disease risks even if the posture is correct.

[0003] Existing studies have shown that it is feasible to use machine learning models to identify worker postures and activities from motion data captured by wearable inertial measurement units (IMUs). With the advancement of machine learning technology, especially the application of neural network models, the classification accuracy of worker posture recognition has been significantly improved. However, the model training process still faces the challenge of high data collection costs. In addition, since workers are highly dynamic during the construction process and are greatly affected by the location environment, relying solely on posture information for identification and ignoring environmental factors may lead to the omission of musculoskeletal disease risks. For example, even if workers maintain correct postures when working in high humidity or low temperature environments, long-term exposure to these conditions may still increase the burden on the musculoskeletal system and lead to work-related injuries or chronic injuries. In addition, current research pays less attention to the identification of monotonous, mechanical cyclic operations, while maintaining correct postures for a long time but performing monotonous, mechanical cyclic operations may also cause musculoskeletal diseases. Therefore, in the current production and construction, how to comprehensively identify the ergonomic risks of workers is still a difficult problem that needs to be solved. Summary of the invention

[0004] In response to the above defects or improvement needs of the prior art, the present invention provides a worker ergonomic risk analysis method and system based on digital twins to solve the problem that the monotonous and mechanical cyclic operations of workers are not considered when conducting risk analysis.

[0005] To achieve the above object, according to one aspect of the present invention, a method for analyzing worker ergonomic risk based on digital twins is provided, and the method comprises the following steps:

[0006] Acquire the worker's spatiotemporal trajectory and motion sequence data; convert the worker's spatiotemporal trajectory and motion sequence data into a region transfer network and a posture transformation network; calculate the number of nodes, average node strength and transfer degree of the region transfer network and the posture transformation network respectively;

[0007] The number of nodes, average node strength and transfer degree of the regional transfer network and the posture transformation network are used to calculate the ergonomic risk associated with a single action and the ergonomic risk associated with a mechanical cycle action, respectively, so as to achieve ergonomic risk analysis.

[0008] Further preferably, before converting the space-time trajectory into the region transfer network and the posture transformation network, the space-time trajectory is converted into a space-time network according to the following steps:

[0009] Discretize the space-time trajectory according to preset time intervals to obtain the position of the worker at each moment after discretization;

[0010] Divide the workers' activity range into regions to obtain the region to which the position at each moment belongs, as well as the transfer relationship between regions at different moments;

[0011] The regions are regarded as nodes, and the transfer relations between the regions are regarded as the edges connecting the nodes, so as to form a network, that is, the space-time trajectory is converted into a space unit transfer network.

[0012] Further preferably, the transfer relationship between the regions includes transfer between different regions and self-transfer within the same region.

[0013] Further preferably, the converting the spatiotemporal trajectory into the regional transfer network is performed according to the following steps:

[0014] The nodes with the same region in the spatial unit network are merged to form a regional node, and the regional transfer relationship on the node before the merger is retained and converted into directed connection lines between the regional nodes, so as to form a new network, namely the regional transfer network, in which the number of directed connection lines on the regional node pointing to the current regional node is used as the weight of the current regional node.

[0015] Further preferably, the step of converting the spatiotemporal trajectory into a posture transformation network is performed according to the following steps:

[0016] In the spatial unit transfer network, for nodes in the same area at different times, determine the posture corresponding to each time, as well as the conversion relationship between postures at different times;

[0017] Nodes with the same posture are merged into one posture node, and the posture conversion relationship on each node before the merger is retained and converted into directed connection lines connecting the posture nodes, so as to form a posture transformation network, wherein the number of wired connection lines on the posture node pointing to the current posture node is used as the weight of the current posture node.

[0018] Further preferably, the formula of the ergonomic risk associated with the single action is as follows:

[0019]

[0020] Among them, Risk 单调 is the ergonomic risk associated with a single action, N R , are the number of nodes in the regional transfer network, the weight and risk value of the i-th node, is the number of posture transformation networks in the i-th node of the region transfer network, N P is the number of nodes in the posture transformation network, and are the node weight and node risk value of the nth node in the jth posture transformation network in the i-th node of the regional transfer network, respectively. i is the node number in the regional transfer network, j is the jth posture transformation network in a certain region, and n is the node number in the posture transformation network.

[0021] Further preferably, the formula for the ergonomic risk associated with the mechanical cyclic action is as follows:

[0022]

[0023] Among them, Risk 循环 are ergonomic risks associated with mechanical cyclic movements, and are the transfer degree, average node strength and number of nodes of the jth posture transformation network in the i-th node of the regional transfer network, N R , are the number of nodes in the regional transfer network, the weight and risk value of the i-th node, is the number of attitude change networks in the i-th node of the regional transfer network; i is the node number in the regional transfer network, and j is the attitude change network number in a certain region.

[0024] Further preferably, the average node strength and transfer degree The formula is as follows:

[0025]

[0026] Where N is the number of nodes, s n is the node weight of the nth node, k n is the number of adjacent nodes of the nth node, t n is the number of edges connecting the adjacent nodes of the nth node.

[0027] According to another aspect of the present invention, a worker ergonomic risk analysis system based on digital twin is provided, the system comprising an actuator for executing the above-mentioned worker ergonomic risk analysis method based on digital twin.

[0028] In general, the above technical solutions conceived by the present invention have the following beneficial effects compared with the prior art:

[0029] 1. The present invention converts the space-time trajectory into a regional transfer network and a posture transformation network. At the same time, when calculating the risk value, two risk calculation methods are used: the ergonomic risk related to a single action and the risk calculation method related to a mechanical cycle action. Both the regional transfer network and the posture transformation network are taken into consideration. The influence of workers' postures, environmental factors and mechanical cycle operations are comprehensively considered to construct a comprehensive worker risk identification model to achieve accurate risk monitoring and management.

[0030] 2. The present invention uses two risk calculation methods, monotonous and cyclic, to independently analyze the health risks caused by workers' awkward postures and mechanical cyclic operations. Monotonous calculation is suitable for static tasks that require maintaining the same posture for a long time to evaluate the risk of fatigue accumulation; cyclic calculation is aimed at highly repetitive mechanical operations to evaluate the fatigue risk of the musculoskeletal system. This dual calculation method can accurately identify and quantify health hazards in different work tasks, optimize the working environment and work flow, and effectively reduce the probability of occupational diseases among workers, thereby improving work safety and health protection.

[0031] 3. The conversion method of the regional transfer network and the posture transformation network provided by the present invention comprehensively considers the correspondence between the worker's posture and the regional environment by constructing a nested network of regional transfer and posture transformation, thereby further considering the environmental impact in the worker's posture risk assessment and improving the accuracy and comprehensiveness of risk analysis.

[0032] 4. The present invention builds a digital twin model of the construction site and sets up an offline simulation mode and a real-time monitoring mode. The offline simulation mode facilitates the collection of training data sets for the workers' posture recognition neural network model, reducing labor costs and time expenditure; the real-time monitoring mode provides data centralization function, which promotes the extraction of scene interaction semantics and centralized data processing and display. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1is a flow chart of a worker ergonomic risk analysis method based on digital twins constructed according to a preferred embodiment of the present invention;

[0034] Figure 2 is a schematic diagram of a space-time trajectory constructed according to a preferred embodiment of the present invention;

[0035] Figure 3 It is a schematic diagram of the structure of converting the space-time trajectory into a region transfer network and a posture transformation network constructed according to the preferred embodiment of the present invention. DETAILED DESCRIPTION

[0036] In order to make the purpose, technical scheme and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0037] A worker ergonomic risk analysis method based on digital twins formulates differentiated safety analysis strategies for different movement patterns of different workers, providing an important reference for adaptive ergonomic risk analysis. The analysis of this study includes three steps: multi-scale spatiotemporal network construction and conversion, and metric extraction. The method specifically includes the following steps:

[0038] (1) Use the digital twin model to obtain the workers’ spatiotemporal trajectory on the construction site and identify their position and posture in the spatiotemporal trajectory.

[0039] Build a digital twin model of the construction site: Convert the 3D models of workers, buildings, and construction equipment into .FBX format and import them into Unity software to build a digital twin model of the construction site.

[0040] Configure the offline simulation mode and real-time monitoring mode of the construction site digital twin model:

[0041] Offline simulation mode: bind the skeleton models of typical construction postures and awkward postures to the worker model, configure the posture state machine, set the corresponding semantic labels for each posture, and realize the switching control between different construction postures of the worker model. Then control the worker model through script settings. Based on this mode, the skeleton node data and corresponding posture semantics of each posture can be output by controlling the movement of the worker model to generate a virtual data set for training the worker posture recognition neural network.

[0042] Real-time monitoring mode: Wearable inertial motion capture equipment and UBW equipment are used to collect worker posture and trajectory data respectively, and data communication between the data collection equipment and the worker model is established to realize the mapping of the worker's motion trajectory to the worker model. The construction area is divided in the digital twin model of the construction site, and collision boxes are set and the area information is marked. The area information where the worker is located is triggered by the spatial intersection of the collision box of the worker model itself and the collision box of a certain area in the digital twin model of the construction site. With the help of the trained worker posture recognition neural network, the posture of the worker model is recognized in real time to obtain the posture information of the worker at every moment.

[0043] Based on this model, the workers' spatiotemporal trajectory and motion sequence data, as well as the area and posture information at each moment can be obtained in real time from the digital twin model of the construction site.

[0044] (2) Convert the spatiotemporal trajectories in the workers’ digital twin data into a spatial unit transfer network. First, select an appropriate unit grid size to divide the construction site into multiple spatial units, assign corresponding semantic information to each cell, and mark the functional attributes of different areas. Then, by tracking the movement paths of workers between different units in the construction site, obtain the time series data of the workers’ movement trajectories. Subsequently, use the time series network (chronnet) technology to convert these time series into complex network models for analysis.

[0045] In order to further explore the regional scale transfer law and attitude scale transformation characteristics of workers during the construction process, the spatial unit transfer network is further converted into a regional transfer network and an attitude transformation network;

[0046] (3) By analyzing the network structure and its related measurement methods, the potential nonlinear risk patterns in workers' behaviors can be identified. In this study, the risk characteristics are quantitatively evaluated based on the complexity and compactness of the network, and finally the ergonomic risks caused by monotonous and mechanical cycle operations under different risk environments are identified through risk intensity weighted calculation.

[0047] Specifically, the transformation of spatiotemporal trajectory data into spatiotemporal networks mainly uses the time series network to transform the time series into a complex network. The specific steps are:

[0048] First, by selecting an appropriate time step and the spatial unit grid coordinates corresponding to each time point, the time series X = {r1, r2, ..., r T}, where T is the total length of the sequence. The i-th (i=1, 2, ..., T) time point r in the sequence i By event timestamp t i , location information i = {La i ,Loi}、Area information j (j=1,2,...,M), action information k (k=1,2,...,N), where M and N are the total number of region categories and action categories respectively.

[0049] Afterwards, according to the spatial position of the trajectory points in the spatial network, if two consecutive time points r i (Located in the unit (La i ,Lo i ) and r i+1 (Located in the unit (La i+1 ,Lo i+1 )) appears continuously within a 1s time window, then from node v i to v j Create a directed edge with weight 1. Node v i and v j Indicates the ID of the unit where two consecutive trajectory points are located. If two nodes are located in the same unit, the node forms a self-loop. This process is repeated until the time length is T. The edge weight between nodes is the number of transfers between adjacent units or within the unit itself. Finally, a weighted directed complex network with self-loops is constructed - a spatial unit transfer network.

[0050] Specifically, the spatial unit transfer network is converted into a regional transfer network. The specific steps are as follows: different spatial units may be in the same region, so the nodes with the same region in the spatial unit transfer network are merged into the same node to generate a regional transfer network. In the regional transfer network, the network nodes are converted into different regions, and the edges between the nodes indicate the existence of regional transfer relationships. The risk values ​​of nodes in different regions are set artificially. The edge weight is the sum of the edge weights of the same regional transfer direction in the spatial unit transfer network, reflecting the number of transfers between the same or different regions. The newly generated network is still a weighted directed complex network with self-loops.

[0051] The spatial unit transfer network is converted into a posture transformation network. The specific steps are as follows: within an area, each worker will have a series of posture changes as the project progresses. Therefore, within an area, the nodes with the same posture in the multiple spatial unit transfer networks contained in it are merged into the same node to generate a posture transformation network. In the posture transformation network, the network nodes are converted into different postures, and the edges between the nodes indicate the existence of posture transformation relationships. According to the ergonomic risk level of each posture, the network nodes are assigned corresponding risk values. The edge weight is the sum of the edge weights of the same posture transfer direction in the spatial unit transfer network, reflecting the number of transfers between the same or different postures. The newly generated network is also a weighted directed complex network with self-loops. The risk value of each posture node is artificially set in advance.

[0052] Specifically, complexity measures the intensity of workers' transitions between different risk levels in the network. Indicators include the number of nodes and the average strength of the network. Compactness measures the possibility of repeated transitions between different risk levels. Metrics include the average network clustering coefficient and network modularity. The corresponding calculation formula is shown in the following table. In complex network theory, the number of nodes N refers to the number of all vertices in the network. A network with a large number of nodes indicates that: in terms of area transfer, the worker has a tendency to cross more dangerous areas; in terms of posture rotation, the worker pays more attention to avoiding mechanical repetitive movements and has a lower chance of suffering from ergonomic injuries. The average strength of the node <s> is the average risk strength of all nodes in the network. The higher the value, the more likely the worker is to cross high-risk areas and use high-risk construction postures.

[0053] The higher the transfer degree T of the network, the tighter the network is and the higher the degree of interconnection between different nodes.

[0054]

[0055] Where N is the number of nodes; s i is the weight of the i-th node, that is, the sum of the weights of all edges pointing to the node; k i is the number of adjacent nodes of the i-th node; t i is the number of edges connecting the adjacent nodes of the i-th node.

[0056] Specifically, the risk intensity weighted calculation is as follows:

[0057]

[0058] Among them, N R , They are the number of nodes, node weights and node risk values ​​of the regional transfer network; is the number of posture transformation networks in the i-th node of the region transfer network; N P is the number of nodes in the posture transformation network; are the node weight and node risk value of the jth posture transformation network in the i-th node of the regional transfer network, respectively; and are the transfer degree, average node weight and number of nodes of the jth posture transformation network in the i-th node of the regional transfer network, respectively.

[0059] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A worker ergonomic risk analysis method based on digital twins, characterized in that: The method comprises the following steps: Acquire the worker's spatiotemporal trajectory and motion sequence data; convert the worker's spatiotemporal trajectory and motion sequence data into a region transfer network and a posture transformation network; calculate the number of nodes, average node strength and transfer degree of the region transfer network and the posture transformation network respectively; The number of nodes, average node strength and transfer degree of the regional transfer network and the posture transformation network are used to calculate the ergonomic risk associated with a single action and the ergonomic risk associated with a mechanical cycle action, respectively, so as to achieve ergonomic risk analysis.

2. A worker ergonomic risk analysis method based on digital twins as claimed in claim 1, characterized in that: The worker's spatiotemporal trajectory and action sequence data are obtained in the construction site digital twin model; before converting the spatiotemporal trajectory into a region transfer network and a posture transformation network, the spatiotemporal trajectory is converted into a spatiotemporal network according to the following steps: Discretize the space-time trajectory according to preset time intervals to obtain the position of the worker at each moment after discretization; Divide the workers' activity range into regions to obtain the region to which the position at each moment belongs, as well as the transfer relationship between regions at different moments; The regions are regarded as nodes, and the transfer relations between the regions are regarded as the edges connecting the nodes, so as to form a network, that is, the space-time trajectory is converted into a space unit transfer network.

3. A worker ergonomic risk analysis method based on digital twins as claimed in claim 2, characterized in that: The transfer relationship between the regions includes transfer between different regions and self-transfer within the same region.

4. A worker ergonomic risk analysis method based on digital twins as claimed in claim 3, characterized in that: The conversion of the spatiotemporal trajectory into a regional transfer network is performed according to the following steps: The nodes with the same region in the spatial unit network are merged to form a regional node, and the regional transfer relationship on the node before the merger is retained and converted into directed connection lines between the regional nodes, so as to form a new network, namely the regional transfer network, in which the number of directed connection lines on the regional node pointing to the current regional node is used as the weight of the current regional node.

5. A worker ergonomic risk analysis method based on digital twins as claimed in claim 4, characterized in that: The transformation of the spatiotemporal trajectory into a posture transformation network is performed according to the following steps: In the spatial unit transfer network, for nodes in the same area at different times, determine the posture corresponding to each time, as well as the conversion relationship between postures at different times; Nodes with the same posture are merged into one posture node, and the posture conversion relationship on each node before the merger is retained and converted into directed connection lines connecting the posture nodes, so as to form a posture transformation network, wherein the number of wired connection lines on the posture node pointing to the current posture node is used as the weight of the current posture node.

6. A worker ergonomic risk analysis method based on digital twins as claimed in claim 5, characterized in that: The formula for the ergonomic risk associated with a single action is as follows: Among them, Risk 单调 is the ergonomic risk associated with a single action, are the number of nodes in the regional transfer network, the weight and risk value of the i-th node, is the number of posture transformation networks in the i-th node of the region transfer network, N P is the number of nodes in the posture transformation network, and are the node weight and node risk value of the nth node in the jth posture transformation network in the i-th node of the regional transfer network, respectively. i is the node number in the regional transfer network, j is the jth posture transformation network in a certain region, and n is the node number in the posture transformation network.

7. A worker ergonomic risk analysis method based on digital twins as claimed in claim 1, characterized in that: The formula for the ergonomic risk associated with the mechanical cycle action is as follows: Among them, Risk 循环 are ergonomic risks associated with mechanical cyclic movements, and are the transfer degree, average node strength and number of nodes of the jth posture transformation network in the i-th node of the regional transfer network, are the number of nodes in the regional transfer network, the weight and risk value of the i-th node, is the number of attitude change networks in the i-th node of the regional transfer network; i is the node number in the regional transfer network, and j is the attitude change network number in a certain region.

8. A worker ergonomic risk analysis method based on digital twins as claimed in claim 6 or 7, characterized in that: The average node strength and transfer degree The formula is as follows: Where N is the number of nodes, s n is the node weight of the nth node, k n is the number of adjacent nodes of the nth node, t n is the number of edges connecting the adjacent nodes of the nth node.

9. A worker ergonomic risk analysis system based on digital twins, characterized in that: The system includes an actuator, which is used to execute a worker ergonomic risk analysis method based on digital twins as described in any one of claims 1-8.

Images