Construction personnel safety monitoring and dispatching system based on internet of things high-precision positioning

Abstract

This invention relates to the fields of IoT intelligent safety monitoring and BeiDou positioning technology, specifically disclosing a construction worker safety monitoring and dispatching system based on IoT high-precision positioning. The system includes a high-precision positioning terminal, a multimodal physiological sensing device, a behavioral feature analysis server, a psychological state assessment engine, and a safety dispatching decision platform. By fusing BeiDou centimeter-level positioning with physiological indicators, it extracts spatiotemporal-physiological features, constructs a stress-behavior mapping model to identify psychological states such as fatigue and anxiety, and generates a psychological stress heatmap. Based on this, the safety dispatching decision platform dynamically adjusts task allocation, triggers tiered early warnings, and pushes intervention instructions. Through the above technical solution, this invention achieves quantitative perception and proactive intervention of psychological risks among construction workers, improves the foresight and humanization of safety management, and reduces safety accidents caused by psychological factors.

Description

Technical Field

[0001] This invention belongs to the field of IoT intelligent safety monitoring and BeiDou positioning technology, specifically involving a construction worker safety monitoring and dispatch system based on IoT high-precision positioning. Background Technology

[0002] With the widespread application of IoT technology in smart construction sites, personnel safety monitoring at construction sites has become a key aspect of reducing accident rates and improving management efficiency. Traditional safety management models are gradually transforming towards digitalization and intelligence, utilizing various sensing technologies to conduct real-time dynamic monitoring of the construction environment and personnel activities. In large-scale and complex infrastructure projects, the dynamic changes in the construction environment and the high frequency of personnel movement place higher demands on the depth of perception, accuracy of early warning, and comprehensiveness of risk assessment of safety monitoring systems.

[0003] Safety dispatch systems based on high-precision positioning technology are a current research focus in the industry, aiming to achieve precise control of work areas by acquiring the real-time geospatial coordinates of construction personnel. High-precision positioning technology primarily relies on the BeiDou satellite navigation system or other wireless sensor networks, combined with logical judgment methods such as electronic fences, to automatically identify and issue alerts for personnel's unauthorized crossing of areas, failure to wear safety helmets, or static gatherings. Its core objective is to map the physical state of the construction site through digital twin technology, ensuring that the work process complies with preset physical safety regulations.

[0004] Existing technologies typically focus only on monitoring overt physical violations, lacking a deep understanding of the physiological and psychological states of construction workers, making it difficult to effectively quantify hidden safety risks. Traditional systems, when processing location data, often only focus on macroscopic positional shifts, neglecting the psychological fluctuations inherent in microscopic movement trajectories, such as hesitation and abnormal path curvature. This results in insufficient ability to identify deeper causes such as fatigue, panic, or inattention. The lack of a deep fusion mechanism between single-dimensional spatial coordinate data and multi-source heterogeneous physiological data makes it difficult to construct accurate behavioral risk prediction models, causing safety monitoring to lag behind in addressing non-linear human error risks.

[0005] Therefore, a construction worker safety monitoring and dispatching system based on high-precision positioning via the Internet of Things is needed. Summary of the Invention

[0006] The purpose of this invention is to provide a construction worker safety monitoring and dispatch system based on high-precision positioning via the Internet of Things, which can solve the problem of the difficulty in quantifying hidden safety hazards in the aforementioned background technology.

[0007] To achieve the above objectives, the technical solution adopted by this invention is: a construction worker safety monitoring and dispatching system based on high-precision positioning via the Internet of Things, comprising a high-precision positioning terminal, a multimodal physiological sensing device, a behavioral feature analysis server, a psychological state assessment engine, and a safety dispatching decision platform. The high-precision positioning terminal is configured to acquire the centimeter-level spatial coordinates of construction workers at the construction site in real time through the BeiDou satellite navigation system, and continuously record their spatiotemporal trajectory data; The multimodal physiological sensing device is integrated into the intelligent safety equipment worn by construction workers. It is configured to synchronously collect physiological indicators such as heart rate variability, skin conductance response, and body surface temperature, and upload the physiological indicators to the behavioral feature analysis server after binding them with corresponding timestamps. The behavior feature analysis server receives spatiotemporal trajectory data from a high-precision positioning terminal and physiological index data from a multimodal physiological sensing device, configures micro-motion features for extracting the trajectory, including path tortuosity, local wandering frequency, regional dwell time, and movement speed fluctuation, and establishes a spatiotemporal-physiological fusion dataset. The psychological state assessment engine constructs a stress-behavior mapping model based on the principles of cognitive psychology. It is configured to input the spatiotemporal-physiological fusion dataset into the stress-behavior mapping model constructed based on the principles of cognitive psychology, identify the current psychological state category of construction workers, including fatigue, anxiety, inattention or emotional panic, and generate an individualized psychological stress heat map. The safety scheduling decision platform is connected to the psychological state assessment engine and is configured to dynamically adjust the allocation of construction tasks based on the psychological stress heat map, trigger a graded early warning mechanism, and push comprehensive scheduling instructions to on-site management personnel, including personnel identity, location, psychological state, and suggested intervention measures.

[0008] Preferably, the high-precision positioning terminal uses a dual-frequency positioning chip that supports BeiDou-3 short message communication, which can maintain continuous positioning capability in tunnels, deep foundation pits or areas shielded by steel structures without public network coverage, and performs noise reduction and smoothing processing on the original trajectory through the edge computing unit to ensure the spatiotemporal consistency of the trajectory data.

[0009] Preferably, the multimodal physiological sensing device has a built-in adaptive filtering algorithm that can dynamically calibrate the physiological signal baseline based on vibration, temperature and humidity changes in the construction environment, eliminate physiological fluctuation interference caused by non-psychological factors, and improve the signal-to-noise ratio of heart rate and skin conductance response data.

[0010] Preferably, the behavior feature analysis server is equipped with a trajectory semantic parsing module, which can transform the original coordinate sequence into activity segments with behavioral semantics, including "normal inspection", "equipment operation", "long-term stillness" or "aimless wandering", and assign differentiated weights to trajectory features in combination with activity type.

[0011] Preferably, the stress-behavior mapping model adopts a multi-layer neural network architecture. Its training process integrates a large amount of historical accident precursor data and psychological experimental labeled samples. It can learn the non-linear correlation between physiological indicator mutations and trajectory abnormalities, and supports online incremental learning to adapt to individual differences in different types of work, ages and working environments.

[0012] Preferably, the psychological state assessment engine is also equipped with a group stress propagation analysis unit, which can identify the emotional contagion effect of high-stress individuals on their surrounding colleagues based on the spatial proximity and interaction frequency among people, and predict the potential trend of group risk spread.

[0013] Preferably, the safety scheduling decision platform integrates a task reassignment optimizer. When it detects that a construction worker is in a high-risk psychological state, it automatically removes the worker from the high-risk work position and assigns a nearby worker with a stable psychological state to take over. At the same time, it links the on-site broadcasting system to play soothing prompts to alleviate tension.

[0014] Preferably, the system also includes a digital twin visualization interface, configured to overlay and display the real-time location, psychological stress level, and historical trajectory playback of each construction worker on a three-dimensional construction site model, allowing managers to intuitively grasp the psychological safety status of all personnel on site through color coding.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. The construction worker safety monitoring and dispatching system based on high-precision positioning of the Internet of Things provided by this invention breaks through the limitation of traditional safety monitoring that only focuses on physical violations. For the first time, it deeply integrates cognitive psychology theory with high-precision spatiotemporal trajectory mining, realizing the quantitative perception and proactive intervention of the hidden psychological risks of construction workers.

[0016] 2. By simultaneously integrating BeiDou centimeter-level positioning data with multimodal physiological indicators, the system constructs individual behavioral profiles. This not only identifies "where a person is" but also determines "what psychological state a person is in," capturing key triggers such as fatigue, panic, or inattention before an accident occurs. By introducing a stress-behavior mapping model, the system can generate dynamic psychological stress maps, supporting precise tiered early warnings and intelligent task scheduling, thus improving the foresight and humanization of safety management.

[0017] 3. The system has the ability to adapt to the environment and learn from individual differences, making it suitable for complex and ever-changing construction sites. It reduces operational errors and safety accidents caused by human psychological factors, and provides a brand-new safety management paradigm for smart construction site construction. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the overall technical solution architecture according to the present invention; Figure 2 This is a schematic diagram illustrating the core principle framework of the stress-behavior mapping model based on cognitive psychology according to the present invention; Figure 3 This is a flowchart illustrating the logical process of spatiotemporal trajectory feature extraction and multimodal physiological index fusion according to the present invention. Figure 4 This is a flowchart illustrating the logical process framework for dynamic task reallocation and hierarchical early warning in the security scheduling decision platform according to the present invention. Figure 5 This is a schematic diagram of the multi-level interaction relationship and data flow between the group pressure propagation analysis and the visualization of individual psychological states in this invention. Detailed Implementation

[0019] Example 1: Please refer to the appendix Figure 1 To be continued Figure 5 To make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to specific embodiments.

[0020] The construction worker safety monitoring and dispatch system based on IoT high-precision positioning includes a high-precision positioning terminal, a multimodal physiological sensing device, a behavioral feature analysis server, a psychological state assessment engine, and a safety dispatch decision platform. The high-precision positioning terminal is used to obtain the spatial coordinates of construction workers within the construction site through the BeiDou satellite navigation system and record spatiotemporal trajectory data. The high-precision positioning terminal is constructed as a wearable smart device, for example, integrated into a safety helmet, work badge, or wristband. The high-precision positioning terminal integrates a high-performance BeiDou-3 dual-frequency positioning chip, which can simultaneously receive satellite signals from multiple frequency bands and achieve centimeter-level positioning accuracy through a combination of carrier phase observations and pseudorange observations.

[0021] To cope with the complex electromagnetic environment at construction sites and the obstruction of satellite signals by tall buildings and lifting machinery, the high-precision positioning terminal is also equipped with an inertial navigation unit, including a three-axis accelerometer and a three-axis gyroscope. When the satellite signal is briefly interrupted due to obstruction, the high-precision positioning terminal uses an inertial accumulation algorithm to perform dead reckoning, ensuring the continuity of spatiotemporal trajectory data.

[0022] The high-precision positioning terminal has a built-in edge computing module configured to preprocess the raw positioning data. The edge computing module uses a Kalman filter algorithm to denoise the raw coordinate sequence, eliminating positioning drift points caused by multipath effects. The edge computing module also has a trajectory compression function, reducing the data sampling and reporting frequency when the person is stationary or moving at a constant speed in a straight line, thereby saving system bandwidth and terminal power consumption.

[0023] The high-precision positioning terminal also supports the short message communication function unique to Beidou-3. In the event of a ground mobile network failure caused by a tunnel, deep foundation pit, or natural disaster, it can send key location information and alarm signals to the management terminal via satellite link.

[0024] The multimodal physiological sensing device is used to synchronously collect the physiological indicators of construction workers and upload them to a behavioral feature analysis server after binding the physiological indicators with corresponding timestamps. The multimodal physiological sensing device includes a heart rate variability sensing unit, a skin conductance response sensing unit, and a body surface temperature sensing unit. The heart rate variability sensing unit uses photoplethysmography (PPG) to capture changes in blood flow within the skin's microvessels through a highly sensitive photoelectric sensor, and then converts these changes into digital pulse signals via an internal signal amplifier and analog-to-digital converter.

[0025] The processor within the multimodal physiological sensing device analyzes pulse signals in real time, calculating the time difference between adjacent heartbeat cycles, i.e., heart rate variability. Heart rate variability is a key parameter for assessing autonomic nervous system activity and psychological stress. The skin conductance sensing unit measures minute changes in skin conductivity through two dry electrodes in contact with the skin. Skin conductance is regulated by sweat gland activity and is a sensitive indicator of emotional arousal and psychological stress.

[0026] The multimodal physiological sensing device incorporates an adaptive filtering algorithm, which dynamically adjusts filtering parameters based on the intensity of the construction worker's activity to filter out artifacts caused by strenuous physical activity. The collected physiological indicators are strictly synchronized with the timestamps generated by the high-precision positioning terminal to ensure precise alignment of spatial location and physiological state on the timeline, generating a physiological data packet with spatiotemporal attributes, which is then transmitted to the data aggregation unit via Bluetooth Low Energy or a wireless LAN.

[0027] The behavioral feature analysis server is used to extract microscopic motion features of trajectories and establish a spatiotemporal-physiological fusion dataset. The behavioral feature analysis server is configured as a cluster system with high-performance parallel processing capabilities, capable of simultaneously processing concurrent data streams from thousands of construction workers. Specifically, the behavioral feature analysis server includes a trajectory feature extraction module, which is configured to calculate path tortuosity from a continuous coordinate sequence. The path tortuosity is defined as the ratio between the total length of the actual path traveled by the person and the straight-line distance from the starting point to the ending point within a specific sliding time window.

[0028] If the ratio between straight-line distances is much greater than 1, it indicates that the person is wandering or loitering without a clear purpose. The trajectory feature extraction module is also configured to calculate the local loitering frequency, that is, the number of times the person moves back and forth within a preset diameter range. The module also records the duration of stay in the area and the fluctuation of movement speed. The fluctuation of movement speed is measured by calculating the standard deviation of the speed, and speed fluctuations are usually related to the person's hesitation, confusion, or physical fatigue.

[0029] The behavioral feature analysis server also includes a trajectory semantic parsing module, which maps the original coordinate sequence to a pre-constructed geographic information model of the construction site. Through joint analysis of location and time, the trajectory semantic parsing module marks trajectory segments as specific behavioral semantics, such as equipment operation behavior on the work surface, movement behavior on the inspection route, or pausing behavior in the rest area. Finally, the behavioral feature analysis server performs multi-dimensional feature vectorization processing on the extracted micro-motion features and corresponding multimodal physiological indicators to construct a spatiotemporal-physiological fusion dataset, which serves as the input source for subsequent psychological state determination.

[0030] The psychological state assessment engine is used to identify the psychological state categories of construction workers and generate a psychological stress heatmap. The engine integrates a stress-behavior mapping model based on cognitive psychology principles. This model employs a multi-layer deep neural network architecture, with its input layer receiving feature vectors from the spatiotemporal-physiological fusion dataset.

[0031] The model's hidden layer contains multiple nonlinear transformation units to capture the deep nonlinear correlation between physiological indicators (such as a sudden increase in skin conductivity) and trajectory anomalies (such as increased path tortuosity). During model training, a large amount of historical task data and manually labeled psychological state samples are incorporated, enabling the model to learn the differences in characteristic expressions of different individuals under varying intensities of stress. The psychological state assessment engine is configured to identify various negative psychological states, including but not limited to fatigue, anxiety, inattention, or panic.

[0032] For example, when heart rate variability decreases while path tortuosity increases, the model outputs a higher anxiety probability value. The psychological state assessment engine also features group stress propagation analysis. Based on a social force model, this module analyzes spatial distance, interaction frequency, and the evolution trend of individual stress values ​​to assess whether high-stress individuals will have a negative emotional contagion effect on surrounding workers. The engine projects the stress scores of all on-duty personnel onto a 3D digital map of the construction site, generating a continuous psychological stress heatmap using kernel density estimation, visually displaying the overall psychological risk distribution at the construction site.

[0033] The safety dispatch decision-making platform is used to dynamically adjust the allocation of construction tasks and trigger early warnings based on the psychological stress heatmap. The platform includes a risk level assessment module, which weights the psychological stress heatmap with the current risk level of the work environment. If the psychological stress value of personnel in a high-risk work area (such as an aerial work platform or a live-line work area) exceeds a first preset threshold, the platform immediately triggers a tiered early warning mechanism.

[0034] The early warning mechanism comprises three levels: Level 1 warning sends alert messages to the mobile terminals of personnel in a high-risk work area (such as an aerial work platform or a live-line work area); Level 2 warning pushes intervention instructions to on-site safety supervisors; and Level 3 warning links the on-site intelligent broadcasting system and lighting system. The safety dispatch decision platform also integrates a task reassignment optimizer, which employs a multi-objective optimization algorithm, prioritizing personnel psychological safety while ensuring project progress. When it detects that an operator in a specific position is in a state of deep fatigue or inattention, the optimizer automatically searches for nearby personnel with stable psychological states and suitable skills, generating a task handover plan.

[0035] The platform also features a digital twin visualization interface, which uses a 3D engine to construct a real-time mirror of the construction site. Managers can view the historical trajectory, real-time psychological state score, and related physiological indicator curves of any worker through interactive operations such as rotation and zoom. The system uses color coding to label personnel at different risk levels; for example, red represents a high-risk stress state, and green represents a stable working state, enabling visualized management and control of hidden risks at the construction site.

[0036] The high-precision positioning terminal employs an encryption authentication protocol during data transmission. When the terminal accesses the system network, it undergoes an authentication process based on an elliptic curve cryptography algorithm. The power management module of the positioning terminal is equipped with adaptive sleep logic. When the built-in accelerometer detects that a person has been stationary for an extended period and their physiological indicators are within the baseline range, it automatically reduces the satellite search frequency of the BeiDou chip to extend battery cycle life. The antenna system of the high-precision positioning terminal uses a multi-feed anti-multipath antenna, which improves the phase center stability of the antenna, further enhancing positioning reliability in construction environments with dense metal components.

[0037] Furthermore, the surface temperature sensing unit in the multimodal physiological sensing device employs a non-contact infrared thermopile sensor, capable of real-time monitoring of changes in a person's skin temperature. When excessively high ambient temperatures cause heat stress in individuals, the abnormal temperature rise signal captured by the surface temperature sensing unit will act as an auxiliary factor inducing psychological anxiety, increasing the proportion of environmental weights in the stress-behavior mapping model. The housing of the multimodal physiological sensing device is made of medical-grade low-sensitivity materials and possesses a high level of waterproof and dustproof capabilities to adapt to the harsh environment of open-air construction site operations.

[0038] Furthermore, the trajectory semantic parsing module in the behavioral feature analysis server utilizes a Hidden Markov Model to model the activity sequences of personnel. The activity sequence model treats the underlying spatial trajectory of the personnel as the observed sequence and the personnel's true behavioral intention as a hidden state. A forward-backward algorithm is used to calculate the most probable sequence of behavioral intentions given a trajectory sequence. This method can effectively distinguish between path detours caused by obstacle avoidance and aimless wandering caused by psychological confusion. The server is configured with a large-capacity distributed storage system for persistently storing historical behavioral profile data of all personnel. By mining long-term historical data, the server can establish personalized behavioral benchmarks for each worker, enabling the system to sensitively capture subtle shifts in individual behavioral patterns.

[0039] The multi-layer neural network in the psychological state assessment engine employs model compression and pruning techniques during deployment to reduce server computing resource consumption and ensure the real-time nature of assessment results. The engine is also equipped with an incremental learning interface, enabling online updates to the neural network's weight parameters upon acquiring new accident case data or expert feedback, thus achieving self-evolution of risk prediction capabilities. In the analysis of group pressure propagation, the system introduces a dynamic graph convolutional network, abstracting work teams at the construction site as nodes and edges in a graph structure. By extracting the feature evolution patterns within the spatial neighborhood, potential risk points of group psychological out-of-control can be identified.

[0040] The safety scheduling decision-making platform comprehensively considers the constraints of the construction plan when executing scheduling decisions. When searching for replacement personnel, the task reassignment optimizer calculates the distance cost from the candidate's current location to the target work point, the candidate's cumulative working time, and their skill proficiency level. The system employs the Pareto optimality principle to find a balance point among multiple conflicting objectives, ensuring the engineering feasibility of the scheduling plan. The platform also supports integration with the on-site access control system, temporarily revoking the access rights of personnel assessed as being in an extremely unsafe psychological state to core high-risk work areas, achieving mandatory safety protection at the physical level.

[0041] Example 2: A construction worker safety monitoring and dispatching system based on IoT high-precision positioning. Its system architecture adopts a heterogeneous model of distributed edge sensing and cloud-based collaborative processing. The construction worker safety monitoring and dispatching system includes distributed high-precision sensing nodes, edge analysis gateways, a cloud-based big data assessment center, and intelligent dispatching terminals.

[0042] The distributed high-precision sensing nodes are deployed on each worker, and their core components include a highly integrated multi-system, multi-frequency GNSS module and a multi-source physiological sensor array. Unlike Embodiment 1, the sensing nodes in this embodiment establish direct communication connections with the cloud via narrowband Internet of Things (NB-IoT), possessing extremely high concurrent connection capabilities. The sensing nodes integrate a primary trajectory filtering unit, which uses moving average filtering to smooth the real-time coordinate stream and calculates basic motion characteristic parameters locally, such as instantaneous displacement vectors and motion intensity.

[0043] The edge analytics gateways are deployed at key nodes on the construction site, such as the top of tower cranes, entrances to office areas, or power distribution rooms. These gateways act as regional data processing centers, responsible for receiving raw physiological electrical signals and trajectory data from surrounding sensing nodes. Each edge analytics gateway is equipped with a high-performance embedded neural network processor, capable of performing complex behavioral feature analysis tasks.

[0044] The edge analytics gateway executes trajectory semantic recognition logic, transforming coordinate data into worker behaviors within specific work areas, such as "walking on scaffolding," "staying inside an elevator," or "carrying materials in a storage area." By performing these data-intensive tasks at the edge, the bandwidth pressure on data transmission to the cloud is significantly reduced, and the response latency from detecting anomalies to triggering alarms is shortened.

[0045] The cloud-based big data assessment center receives processed data aggregated from various edge analytics gateways and combines it with the overall work plan to conduct in-depth psychological risk prediction. The cloud center maintains a dynamic digital twin model of the construction environment, which not only includes static building geometry information but also updates environmental parameters in real time, such as wind speed, noise levels, and light intensity. The psychological state assessment engine within the cloud-based assessment center, based on cognitive load theory, has established a multi-dimensional risk assessment matrix.

[0046] The multidimensional risk assessment matrix consists of three dimensions: a behavioral anomaly index based on trajectory mining, a stress response index based on physiological signs, and a third dimension representing the severity of the environment. By calculating the Euclidean distance between the three-dimensional vectors in the risk space, the system generates a dynamic safety coefficient for each individual. The cloud center also utilizes deep reinforcement learning algorithms to simulate the overall scheduling strategy and find the optimal task allocation path.

[0047] The intelligent dispatch terminal is a high-performance mobile device or a large-size command center display screen equipped for management personnel. The terminal runs interactive 3D visualization control software, which can display dynamic images of the construction site in real-time rendering. Each construction worker has a corresponding 3D image in the digital twin model, and the image's color changes in real-time according to their psychological stress level.

[0048] Managers can use gestures to select specific individuals exhibiting abnormal behavior and view their psychological stress evolution curve over the past 24 hours, as well as a correlation analysis report of multimodal physiological indicators.

[0049] In this distributed architecture, the high-precision sensing nodes are configured with bidirectional feedback capabilities. When the cloud-based big data assessment center identifies a person as being in a high-risk state, it sends instructions to the sensing node worn by the person, which then issues a tactile or voice warning directly to the person via a built-in miniature vibration motor or voice synthesis unit. This closed-loop control mechanism achieves second-level safety intervention without relying on management personnel intervention.

[0050] The edge analytics gateway also possesses offline autonomy. In the event of a temporary interruption of cloud communication, the edge gateway can autonomously issue warnings for serious violations or extreme physiological anomalies within its jurisdiction based on locally stored risk assessment logic. This architecture ensures that the system maintains basic functional integrity even under extreme conditions of network instability.

[0051] The cloud-based big data assessment center also includes a long-term health management archive. By statistically analyzing workers' physiological, psychological, and behavioral data over months or even years, the system can identify individuals with occupational health risks, such as chronic fatigue caused by prolonged high-intensity work, providing construction companies with precise workforce optimization plans and medical examination recommendations. Regarding data privacy protection, the cloud center employs differential privacy technology, protecting the original sensitive information of individual physiological characteristics while conducting risk prediction and analysis on group data.

[0052] The intelligent scheduling terminal integrates a collaborative decision support system into its interactive logic. When large-scale task adjustments occur, the collaborative decision support system automatically lists the potential impact of the scheduling plan on project nodes and provides alternative risk hedging strategies. For example, when a large number of personnel need to stop work and rest due to collective anxiety caused by high temperatures, the system automatically calculates the resource allocation plan for catching up on the schedule, achieving dual optimization of safety and progress.

[0053] This embodiment, through the deep integration of edge computing and cloud big data, not only improves the robustness of the system in complex environments, but also transforms abstract psychological pressure into concrete management instructions through digital twin technology, thereby improving the scientific nature of safety scheduling at the construction site.

[0054] Example 3: A construction worker safety monitoring and dispatching system based on IoT high-precision positioning, specifically adapted for confined spaces such as underground rail transit projects. The system includes a multi-source fusion positioning network, ergonomic smart wearable kits, a synchronous sensing and analysis matrix, and a collaborative early warning command tower.

[0055] In confined spaces (such as tunnels and shield tunnel sections), where satellite signals cannot provide coverage, the multi-source fusion positioning network employs a seamless switching mechanism between ultra-wideband (UWB) positioning technology and BeiDou positioning technology. In open areas such as tunnel entrances, positioning terminals carried by personnel use BeiDou positioning; once personnel enter the tunnel, the terminals automatically switch to ranging communication with UWB base stations pre-deployed on the tunnel walls.

[0056] The positioning terminal integrates a multi-sensor fusion filter, which fuses centimeter-level ranging information from UWB with gait characteristics from microelectromechanical systems (MEMS) inertial sensors, enabling high-frequency, low-drift trajectory tracking in confined spaces.

[0057] The ergonomic smart wearable kit adopts a modular design, including flexible electrophysiological electrodes integrated into the lining of the safety helmet, an infrared temperature sensing module integrated into the cuffs of the work clothes, and a six-axis motion posture sensor integrated into the waist belt. The flexible electrodes can acquire high signal-to-noise ratio electrocardiogram signals with less skin contact pressure, thereby extracting more accurate heart rate variability characteristics.

[0058] The posture sensor at the waist belt can determine whether a person has fallen, collided, or is in an extremely cramped working posture by identifying the torso tilt angle and sudden acceleration changes. This physical posture information is considered an important lateral parameter for assessing psychological panic.

[0059] The synchronous sensing and analysis matrix is ​​a high-performance computing unit deployed in the on-site command center. Employing a streaming computing framework, it can analyze large-scale trajectory stream data from a multi-source fusion positioning network in real time. For the typical "pendulum-like" work trajectory within the tunnel, the analysis matrix defines a series of unique behavioral templates.

[0060] For example, when personnel remain in high-risk areas such as the cutterhead of a tunnel boring machine for more than a set safety threshold, and this is accompanied by irregular changes in breathing rate, the analysis matrix will immediately initiate a cognitive load assessment process. This process calculates the compatibility between the complexity of the personnel's current task and their physiological capacity, outputting a real-time safety redundancy value.

[0061] The collaborative early warning command tower integrates an audio-visual guidance system with a handheld augmented reality (AR) dispatch terminal. In dimly lit and noisy environments like tunnels, traditional voice or screen alarms have limited effectiveness. The collaborative early warning command tower controls color-changing LED light strips laid on the tunnel ceiling, using the direction of color changes to guide high-stress personnel to evacuate to a safe area.

[0062] On-site team leaders can use an AR terminal to scan the work site and see psychological state tags and historical behavior paths superimposed on real personnel on the screen, achieving a "what you see is what you get" safety monitoring mode.

[0063] The audio-visual guidance system is also specially equipped with a gas environment correlation sensing unit. In confined spaces, environmental factors such as oxygen content and carbon monoxide concentration directly affect people's cognitive abilities. The system synchronously inputs real-time environmental monitoring data into the psychological state assessment engine. When environmental indicators deteriorate to a set threshold, the engine automatically increases the sensitivity weight of physiological indicators in the stress-behavior mapping model, realizing the coupled assessment of environmental factors and psychological state.

[0064] The safety dispatch decision-making platform incorporates a "dynamic escape route planning" algorithm into its logic for tunnel engineering. When a sudden emergency occurs (such as water seepage or fire), the platform plans the optimal escape route for each person based on their real-time location and psychological resilience (assessing whether they are in a state of panic paralysis), and guides personnel to evacuate through tactile feedback (vibration in different modes) from smart wearable kits.

[0065] The system also includes an offline analysis subsystem for retrospectively analyzing the psychological stress maps of the previous shift's personnel during shift handover. This analysis identifies work procedures or physical locations that easily induce psychological stress, guiding construction units to optimize and modify the work environment, such as increasing lighting or improving ventilation, thereby reducing safety risks at the source.

[0066] In processing multimodal physiological data, the system incorporates a federated learning mechanism. Because the physiological data of employees from different construction companies is sensitive, federated learning allows multiple construction project teams to jointly train a psychological stress assessment model without exchanging raw, private data. This leverages a broader sample set to improve the model's generalization ability, enabling it to adapt to the psychological response patterns of workers of different ages and nationalities.

[0067] This embodiment solves the problem of traditional security monitoring being "invisible, not deeply perceptible, and not easily responsive" in underground engineering by constructing a comprehensive perception and intervention system within a confined space, and achieves a deep integration of physical space security and psychological space health.

[0068] Example 4: A construction worker safety monitoring and dispatching system based on IoT high-precision positioning, characterized by the introduction of machine vision-based behavior verification and digital twin real-time synchronization technology. The construction worker safety monitoring and dispatching system includes a multispectral visual perception array, an edge visual analysis unit, a trajectory-visual fusion center, and a cross-platform 3D dispatching portal.

[0069] The multispectral visual sensing array is deployed on fixed tower cranes, columns, and mobile inspection robots at the construction site. These cameras not only include the visible light band but also integrate thermal imaging, enabling all-weather, multi-dimensional target capture. The sensing array's task is to capture real-time images of workers' movements, such as limb extension angles, the continuity of tool operation, and subtle changes in facial expressions.

[0070] The edge visual analysis unit is deployed close to the multispectral visual perception array and runs a skeletal point recognition algorithm based on a deep convolutional neural network. This algorithm simplifies the human body into a skeletal model composed of dozens of key points. By analyzing the motion vectors of these skeletal points, it accurately determines the current workload and the standardization of a person's movements. The edge unit also performs facial micro-expression recognition, capturing visual features highly correlated with stress and fatigue, such as frequent blinking and frowning, and compressing these features into low-bandwidth behavioral feature codes before sending them to the cloud.

[0071] The trajectory-visual fusion center is the core processing hub of this embodiment. It receives spatial coordinate streams from the BeiDou high-precision positioning terminal and behavioral feature codes from the edge visual analysis unit. The fusion center uses a spatiotemporal consistency constraint algorithm to accurately spatially register the microscopic movements captured by vision with the high-precision trajectory.

[0072] For example, when the location trajectory shows that a person is stationary, the fusion center uses visual data to determine whether they are performing a fine-tuning operation or resting in place. This complementary mechanism of multimodal data eliminates the misjudgments that may occur with a single sensor.

[0073] The psychological state assessment engine within the fusion center incorporates the "attention allocation model" from cognitive psychology. By analyzing the deviation between the worker's gaze projection area (based on visual recognition of head posture) and the actual work point, it quantifies the degree of their attentional distraction.

[0074] The cross-platform 3D dispatch portal employs cross-terminal graphics rendering technology, supporting simultaneous display on the web, mobile app, and large-screen command center. The digital twin model in the portal has a real-time behavior playback function, allowing administrators to click on any person on the map to view an "abnormal behavior slice" (i.e., the visual and trajectory fusion screen that triggered the risk assessment) automatically edited by the system.

[0075] The system also integrates an intelligent voice dispatch assistant, allowing managers to query "the top 5 workers with the highest stress levels in the entire site" or "the work group with the least concentration in the past hour" via voice commands, and initiate group calls for real-time command with one click.

[0076] The safety scheduling decision-making platform has added a "skill-psychology bidirectional adaptation" algorithm to its task redistribution logic. This algorithm assesses each worker's proficiency in a specific job based on historical visual data, and combines this with their current psychological stress level to recommend the most suitable task for their current state. For example, for a highly skilled welder currently experiencing moderate fatigue, the system might suggest switching to less demanding material handling work, while assigning the precision welding task to a more competent replacement.

[0077] This embodiment also includes a holographic early warning and linkage device. Holographic projection equipment is deployed in key monitoring areas of the construction site. When the system determines that personnel in a certain area are in a high-risk psychological state, the holographic projection will project dynamic safety warning signs and evacuation instructions onto the ground in front of them. This invasive visual feedback can directly affect the subconscious of personnel and reduce the probability of human error.

[0078] At the data storage and mining level, the system adopts a non-relational spatiotemporal database, supporting joint indexing of massive visual feature codes and trajectory coordinates. By retrospectively analyzing the psychological state of personnel during the construction period, managers can identify "peak periods of psychological risk" in the project (such as when the project deadline is approaching), and thus intervene in advance to prevent safety accidents by increasing welfare benefits and adjusting work schedules.

[0079] This embodiment fills the gap in the perception of micro-behavior in traditional Internet of Things positioning by introducing the dimension of machine vision, and constructs a comprehensive closed-loop monitoring network from macro position to micro action, from physiological signs to psychological state.

[0080] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A construction worker safety monitoring and dispatching system based on high-precision positioning via the Internet of Things, characterized in that, include: The high-precision positioning terminal is configured to obtain the centimeter-level spatial coordinates of construction workers at the construction site through the BeiDou satellite navigation system and continuously record spatiotemporal trajectory data. A multimodal physiological sensing device is integrated into the intelligent safety equipment worn by construction workers. It is configured to simultaneously collect physiological indicators such as heart rate variability, skin conductance response, and body surface temperature, and bind the physiological indicators to corresponding timestamps. A behavioral feature analysis server is connected to the high-precision positioning terminal and the multimodal physiological sensing device, and is configured to extract microscopic motion features from the spatiotemporal trajectory data, and combine the physiological indicators to construct a spatiotemporal-physiological fusion dataset. A psychological state assessment engine is configured to perform feature recognition on the spatiotemporal-physiological fusion dataset based on a stress-behavior mapping model, determine the psychological state category of construction workers, and generate a psychological stress heat map of the construction site. The safety scheduling decision platform is connected to the psychological state assessment engine and is configured to dynamically adjust the allocation of construction tasks based on the psychological stress heat map. It also triggers a graded early warning mechanism based on the risk level of the psychological state category and pushes scheduling instructions containing intervention measures to the management end.

2. The construction worker safety monitoring and dispatching system based on high-precision positioning via the Internet of Things as described in claim 1, characterized in that: The high-precision positioning terminal integrates a positioning chip that supports BeiDou-3 dual-frequency positioning signals. The positioning chip achieves centimeter-level positioning by combining carrier phase observations and pseudorange observations. The high-precision positioning terminal also includes an inertial navigation unit, which consists of a three-axis accelerometer and a three-axis gyroscope, and is configured to maintain trajectory continuity using dead reckoning algorithms during satellite signal interruptions. The high-precision positioning terminal has a built-in edge computing module, which is equipped with a Kalman filter unit to denoise the original coordinate sequence and remove positioning drift points. The edge computing module also has a trajectory compression function to reduce the data sampling frequency when the person is stationary or in uniform linear motion. The high-precision positioning terminal also has a BeiDou-3 short message communication module, which is used to transmit location information and alarm signals via satellite link in environments where terrestrial mobile networks are unavailable.

3. The construction worker safety monitoring and dispatching system based on high-precision positioning via the Internet of Things as described in claim 2, characterized in that: The multimodal physiological sensing device includes a heart rate variability sensing unit, a skin conductance sensing unit, and a body surface temperature sensing unit. The heart rate variability sensing unit uses a photoplethysmography pulse wave sensor to acquire pulse signals by capturing changes in blood flow in the skin's microvessels and to calculate the time difference between adjacent heartbeat cycles. The skin electrical response sensing unit includes two dry electrodes that contact the skin for measuring changes in skin conductivity caused by sweat gland activity. The body surface temperature sensing unit uses a non-contact infrared thermopile sensor to monitor changes in skin temperature in real time. The multimodal physiological sensing device has a built-in adaptive filter processor, which is configured to dynamically adjust the filter parameters according to the movement intensity of the construction workers to filter out artifact interference caused by physical activity. The collected physiological indicators are sent to the behavioral feature analysis server via Bluetooth Low Energy or Wi-Fi.

4. The construction worker safety monitoring and dispatching system based on high-precision positioning via the Internet of Things as described in claim 3, characterized in that: The behavior feature analysis server includes a trajectory feature extraction module, which is configured to calculate path tortuosity, local wandering frequency, regional dwell time, and movement speed fluctuation within a specific sliding time window. The path tortuosity is defined as the ratio between the total length of the actual path traveled by the person and the straight-line distance from the starting point to the ending point; The local wandering frequency is defined as the number of times a person moves back and forth within a preset diameter range; The movement speed fluctuation is measured by calculating the standard deviation of the speed within the sliding time window; The behavioral feature analysis server is also equipped with a vectorization processing unit, which is used to perform multi-dimensional feature fusion of the extracted micro-motion features and the physiological indicators to generate a spatiotemporal-physiological fusion feature vector representing individual behavioral features.

5. The construction worker safety monitoring and dispatching system based on IoT high-precision positioning according to claim 4, characterized in that: The behavior feature analysis server also includes a trajectory semantic parsing module, which is configured with a behavior recognition unit based on a hidden Markov model. The behavior recognition unit takes the spatial trajectory as the observation sequence and the person's true behavioral intention as the hidden state. It uses a forward-backward algorithm to calculate the behavioral intention sequence under a given trajectory sequence, thereby transforming the original coordinate sequence into an activity segment with behavioral semantics. The activity segments include routine inspections, equipment operation, prolonged periods of stillness, and aimless wandering. The trajectory semantic parsing module assigns differentiated weights to the trajectory features in the spatiotemporal-physiological fusion dataset according to the type of the activity segment, in order to distinguish between path curvature caused by environmental obstacle avoidance and trajectory anomalies caused by psychological factors.

6. The construction worker safety monitoring and dispatching system based on IoT high-precision positioning according to claim 5, characterized in that: The stress-behavior mapping model in the psychological state assessment engine adopts a multi-layer neural network architecture. Its input layer receives the feature vector of the spatiotemporal-physiological fusion dataset, and its hidden layer contains multiple nonlinear transformation units to capture the nonlinear correlation between physiological index mutations and trajectory abnormal features. The psychological state assessment engine is configured to identify fatigue, anxiety, inattention, and panic. The psychological state assessment engine also has an online incremental learning interface, which is used to update the weight parameters of the neural network based on newly acquired accident case data or management feedback results. The psychological state assessment engine employs model compression and pruning techniques to reduce the consumption of server computing resources and ensure the real-time nature of psychological state determination.

7. The construction worker safety monitoring and dispatching system based on IoT high-precision positioning according to claim 6, characterized in that: The psychological state assessment engine is also equipped with a group stress propagation analysis unit. The group stress propagation analysis unit is based on the social force model and dynamic graph convolutional network. It abstracts the work teams at the construction site as nodes and edges in a graph structure. By analyzing the spatial distance between personnel, the frequency of interaction, and the evolution trend of individual psychological stress values, it identifies the emotional contagion effect of high-stress individuals on surrounding workers. The psychological state assessment engine uses kernel density estimation to project the psychological stress scores of all on-duty personnel onto a three-dimensional digital map of the construction site, generating a continuously distributed psychological stress heat map to visually demonstrate the diffusion trend and distribution of psychological risks across the entire site.

8. The construction worker safety monitoring and dispatching system based on high-precision positioning via the Internet of Things as described in claim 7, characterized in that: The safety dispatch decision platform includes a risk level assessment module, which is configured to perform a weighted calculation of the psychological stress heat map and the inherent risk level of the work environment. When the psychological stress value of personnel in the work area exceeds a preset threshold, a three-level early warning mechanism is triggered. The Level 1 warning system sends tactile vibration or voice alerts to the high-precision positioning terminals of relevant personnel. A Level 2 warning involves sending an intervention command to the mobile device of the on-site safety supervisor. The intervention command includes the identity and specific location information of the affected personnel. The Level 3 early warning system will play a soothing alert tone through the intelligent broadcasting system at the construction site and trigger the lighting system to flash warning lights.

9. The construction worker safety monitoring and dispatching system based on high-precision positioning via the Internet of Things as described in claim 8, characterized in that: The safety scheduling decision platform integrates a task reassignment optimizer. The task reassignment optimizer uses a multi-objective optimization algorithm. When it detects that a construction worker is in a high-risk psychological state, it automatically searches for a replacement among the surrounding candidates whose psychological state is stable and whose skill proficiency meets the requirements. The multi-objective optimization algorithm comprehensively considers the distance cost from the candidate personnel to the target work point, the cumulative working time of the candidate personnel, and the constraints of personnel skill level during the search process, and generates a task handover scheme according to the Pareto optimality principle; the safety scheduling decision platform is also linked with the access control system of the construction site, and for personnel whose assessment results are in an extremely unsafe state, the platform temporarily revokes their access rights to enter the core high-risk work area.

10. The construction worker safety monitoring and dispatching system based on high-precision positioning via the Internet of Things as described in claim 9, characterized in that: The system also includes a digital twin visualization interface, which uses a 3D engine to construct a real-time mirror of the construction site and overlays the real-time location, psychological stress level and historical trajectory of the construction workers on the 3D model. The digital twin visualization interface supports color coding to label personnel at different risk levels, with red representing high-risk psychological stress and green representing stable work status. The system adopts a distributed edge perception and cloud collaborative processing architecture. The edge computing module performs trajectory denoising and primary feature extraction tasks at the edge, while the cloud big data assessment center performs long-term psychological risk prediction and historical behavior profile mining. The high-precision positioning terminal is equipped with adaptive sleep logic. When the accelerometer detects that a person has been stationary for a long time and that their physiological indicators are within the baseline range, it automatically reduces the positioning satellite search frequency to extend battery life.

Images