Railway construction man-machine information intelligent management method based on RFID

Abstract

This invention discloses an intelligent management method for human-machine information in railway construction based on RFID, belonging to the field of railway construction safety management technology. The method includes: collecting raw radio frequency signals and environmental data from personnel and machinery tags within the construction area; constructing and updating a digital twin map in real time based on the raw radio frequency signals and environmental data; using this map to adaptively calibrate tag signals to obtain high-precision spatiotemporal trajectories; inputting the high-precision spatiotemporal trajectories into a spatiotemporal relationship inference engine to output quantified safety risk levels; and iteratively optimizing the digital twin map and physical rule model based on the feedback results of risk warnings. This invention achieves accurate calibration and compensation of RFID signals in complex construction environments by constructing a dynamically updated wireless signal digital twin map; and utilizes a spatiotemporal relationship inference engine to fuse multi-dimensional data for forward-looking risk simulation, outputting quantified safety risk levels, thus realizing a shift from passive alarm to proactive warning.

Description

Technical Field

[0001] This invention relates to the field of railway construction safety management technology, and in particular to an intelligent management method for railway construction human-machine information based on RFID. Background Technology

[0002] Current railway construction safety management mainly relies on GPS, video surveillance, and traditional RFID area positioning technology. These methods have significant limitations in complex construction environments: GPS signals are easily blocked and interfered with by multipath effects, resulting in insufficient positioning accuracy; video surveillance has blind spots and is easily constrained by environmental conditions; and conventional RFID technology lacks modeling and compensation mechanisms for dynamically changing environmental factors (such as temperature, humidity, electromagnetic noise, and metal obstacles), resulting in poor radio frequency signal stability, large positioning errors, and an inability to achieve high-precision spatiotemporal trajectory tracking of people and machines. At the same time, existing systems mostly focus on post-event response and lack the ability to proactively extrapolate risks and self-learn and optimize by integrating multi-dimensional information from people, machines, and the environment, making it difficult to meet the management needs of modern intelligent railway construction in terms of the accuracy and real-time performance of safety warnings.

[0003] However, current common solutions have many drawbacks, including: conventional technologies such as RFID, GPS, and video surveillance used in railway construction safety management lack dynamic modeling and signal compensation mechanisms for complex environmental factors, resulting in low positioning accuracy and poor stability; at the same time, existing methods are mostly based on simple threshold judgments, failing to achieve deep integration and spatiotemporal correlation analysis of multi-dimensional information of people, machines, and environment, leading to delayed risk assessment and insufficient foresight; in addition, the systems are generally rigid and unable to self-optimize based on construction progress and feedback results, making it difficult to meet the needs of high-precision, adaptive, and intelligent modern railway construction safety management. Summary of the Invention

[0004] The purpose of this section is to outline some aspects of embodiments of the present invention and to briefly describe some preferred embodiments. Simplifications or omissions may be made in this section, as well as in the abstract and title of this application, to avoid obscuring the purpose of these documents; however, such simplifications or omissions should not be construed as limiting the scope of the invention.

[0005] In view of the problems existing in the above-mentioned RFID-based intelligent management methods for railway construction personnel and machinery information, this invention is proposed.

[0006] Therefore, the purpose of this invention is to provide an intelligent management method for railway construction personnel and machinery information based on RFID. This method is applicable to solving the problems of low positioning accuracy and poor stability caused by the lack of dynamic modeling and signal compensation mechanisms for complex environmental factors currently used in railway construction personnel and machinery safety management. At the same time, existing methods are mostly based on simple threshold judgments and fail to achieve deep integration and spatiotemporal correlation analysis of multi-dimensional information of personnel, machinery and environment, resulting in problems of delayed risk assessment and insufficient predictability.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, embodiments of the present invention provide an intelligent management method for railway construction personnel and machinery information based on RFID. This method includes: collecting raw radio frequency signals and environmental data from personnel and machinery tags within a construction area via a radio frequency identification (RFID) reader network; constructing and updating a digital twin map of the construction environment's wireless signals in real time based on the raw radio frequency signals and environmental data; using this map to adaptively calibrate the tag signals and calculate a high-precision spatiotemporal trajectory of the target; inputting the high-precision spatiotemporal trajectory into a spatiotemporal relationship inference engine; performing deduction based on a built-in physical rule model; and outputting a quantified safety risk level; and based on the feedback results of the risk warning, performing collaborative iterative optimization of the digital twin map and the physical rule model to achieve intelligent management of railway construction personnel and machinery information.

[0008] As a preferred embodiment of the RFID-based intelligent management method for railway construction human-machine information described in this invention, the environmental data includes temperature, humidity, noise intensity, electromagnetic interference level, satellite positioning signal quality, and on-site geometric structure, metal obstacle distribution information, and construction progress plan data extracted from the construction BIM model.

[0009] As a preferred embodiment of the RFID-based intelligent management method for railway construction human-machine information described in this invention, the specific content of constructing and real-time updating the digital twin map of wireless signals in the construction environment is as follows: A training dataset is obtained, which includes the RFID radio frequency signal features of tags collected under multiple sets of different environmental data and their corresponding actual tag location coordinates; a positioning calibration model is trained using the training dataset based on a machine learning algorithm, the positioning calibration model being used to characterize the mapping relationship between RFID signal features, environmental data, and spatial location; the real-time collected RFID signals and environmental data are input into the trained positioning calibration model to calculate the signal propagation parameters of the current environment; based on the signal propagation parameters, the digital twin map representing the wireless signal propagation characteristics of the entire construction area is generated and updated.

[0010] As a preferred embodiment of the RFID-based intelligent management method for railway construction human-machine information described in this invention, the step of adaptively calibrating the tag signal using a digital twin map to calculate the high-precision spatiotemporal trajectory of the target specifically includes the following steps: real-time acquisition of the original radio frequency signal of the tag and current environmental data within the construction area; querying the signal propagation parameters of the corresponding area from the digital twin map based on the current environmental data, including path loss coefficient, multipath attenuation factor, and environmental interference weight; compensating and calibrating the original radio frequency signal based on the queried signal propagation parameters to eliminate the influence of environmental factors on signal measurement; using a weighted trilateration algorithm to calculate the preliminary spatial coordinates of the tag using the calibrated signal strength value; and combining inertial measurement unit data with a Kalman filter algorithm to perform spatiotemporal fusion processing on the preliminary spatial coordinates to generate a smooth and continuous high-precision spatiotemporal trajectory.

[0011] As a preferred embodiment of the RFID-based intelligent management method for railway construction human-machine information described in this invention, the specific process of the spatiotemporal relationship reasoning engine includes: establishing a spatiotemporal relationship matrix of human-machine environment to quantify the spatiotemporal correlation between personnel, machinery, and environmental elements; calculating the probability of potential conflicts based on the spatiotemporal correlation; using a multi-level risk assessment algorithm to generate a quantified safety risk level by combining the spatiotemporal correlation and the probability of conflict; and establishing a dynamic risk threshold adjustment mechanism to adaptively adjust the risk judgment criteria according to the construction progress and environmental changes.

[0012] As a preferred embodiment of the RFID-based intelligent management method for railway construction human-machine information described in this invention, the specific content of the safety risk levels is as follows: High risk level, the determination condition is: the potential conflict probability is greater than or equal to the high risk probability threshold or the Euclidean distance between objects is less than or equal to the high risk distance threshold. If either condition is met, it is determined to be high risk, indicating that a collision or safety accident will occur, and emergency intervention measures must be taken immediately; Medium risk level, the determination condition is: it does not reach the high risk level, and the potential conflict probability is greater than or equal to the medium risk probability threshold and less than the high risk probability threshold, or the Euclidean distance between objects is less than or equal to the medium risk distance threshold. This level indicates that there is a potential dangerous trend, and an early warning should be issued to remind relevant personnel to pay attention; Low risk level, the determination condition is: it does not reach the high risk or medium risk level, that is, the potential conflict probability is less than the medium risk probability threshold and the Euclidean distance between objects is greater than the medium risk distance threshold. This level indicates that the spatiotemporal interaction between targets is weak and they are in a normal working state. The system only needs to keep monitoring.

[0013] As a preferred embodiment of the RFID-based intelligent management method for railway construction human-machine information described in this invention, the collaborative iterative optimization process adopts an adaptive optimization framework based on reinforcement learning, specifically including: establishing an optimization objective function, comprehensively considering multi-dimensional performance indicators such as positioning accuracy, early warning accuracy, and system response time; designing a reward mechanism to dynamically adjust the update frequency of the digital twin map and the parameters of the physical rule model according to the risk early warning effect; adopting a distributed optimization strategy, performing local optimization on edge computing nodes and global model aggregation in the cloud; and establishing a model performance evaluation system to periodically evaluate and provide feedback on the optimization effect.

[0014] Secondly, to further address the aforementioned technical problems, the present invention provides an RFID-based intelligent management system for railway construction personnel and machinery information, comprising: a data acquisition module for collecting raw radio frequency signals and environmental data of personnel and machinery tags within the construction area via a radio frequency identification reader network; a map construction module for constructing and updating a digital twin map of wireless signals in the construction environment in real time based on the raw radio frequency signals and environmental data, and using the map to adaptively calibrate the tag signals to calculate the high-precision spatiotemporal trajectory of the target; a level output module for inputting the high-precision spatiotemporal trajectory into a spatiotemporal relationship inference engine, performing deduction based on a built-in physical rule model, and outputting a quantified safety risk level; and an iterative optimization module for performing collaborative iterative optimization of the digital twin map and the physical rule model based on the feedback results of risk warnings, thereby realizing intelligent management of railway construction personnel and machinery information.

[0015] Thirdly, embodiments of the present invention provide a computer device, including a memory and a processor, wherein the memory stores a computer program, and the computer program, when executed by the processor, implements any step of the intelligent management method for railway construction human-machine information based on RFID as described in the first aspect of the present invention.

[0016] Fourthly, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the RFID-based intelligent management method for railway construction human-machine information as described in the first aspect of the present invention.

[0017] The beneficial effects of this invention are as follows: By constructing a dynamically updated digital twin map of wireless signals, this invention achieves accurate calibration and compensation of RFID signals in complex construction environments, significantly improving the accuracy and stability of spatiotemporal trajectory positioning for personnel and machinery; by using a spatiotemporal relationship inference engine to fuse multidimensional data for forward-looking risk simulation and outputting quantified safety risk levels, it realizes the transformation from passive alarm to proactive early warning; in particular, by adopting a cooperative iterative optimization mechanism based on reinforcement learning, the system can continuously evolve itself based on early warning feedback, continuously improving positioning accuracy, early warning accuracy, and response speed, ultimately forming a highly reliable, self-learning, and adaptive high-precision proactive protection system for railway construction safety. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein: Figure 1 This is a flowchart illustrating the implementation of the present invention in Example 1.

[0019] Figure 2 This is an overall flowchart of the intelligent management method for railway construction human-machine information based on RFID in Example 1. Detailed Implementation

[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0021] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0022] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0023] Example 1 Reference Figure 1 and Figure 2 This is the first embodiment of the present invention, which provides an intelligent management method for railway construction human-machine information based on RFID, including the following steps: S1: Collect raw radio frequency signals and environmental data of personnel and machinery tags within the construction area through a network of radio frequency identification (RFID) readers.

[0024] Preferably, the environmental data includes temperature, humidity, noise intensity, electromagnetic interference level, satellite positioning signal quality, as well as on-site geometry, metal obstacle distribution information, and construction schedule data extracted from the construction BIM model.

[0025] Specifically, environmental data is collected in real time through environmental sensors distributed throughout the construction area.

[0026] Furthermore, the readers are arranged in a honeycomb topology, with the spacing set at 50-100 meters depending on the site environment, ensuring full coverage of the construction area without blind spots. Each reader is equipped with multiple environmental sensors, including temperature sensors, humidity sensors, electromagnetic interference monitors, and noise sensors, which are transmitted to the central processing server via industrial Ethernet.

[0027] Specifically, the collected raw radio frequency signals and environmental data are preprocessed to ensure data quality and consistency, providing reliable input for subsequent processing. The preprocessing includes signal filtering, data cleaning and outlier removal, timestamp alignment, data fusion, and data standardization.

[0028] For example, in a railway tunnel construction scenario, 13 RFID readers with integrated environmental sensors are arranged at 80-meter intervals within a 1000-meter tunnel in a honeycomb topology. Personnel wear RFID wristbands with built-in IMUs, and anti-metal RFID tags are affixed to machinery. The readers collect raw radio frequency signals (including received signal strength, carrier phase, etc.) from personnel and machinery, while the sensors collect environmental data such as temperature and humidity. Simultaneously, information such as tunnel cross-section structure and construction progress, as well as IMU motion data, are extracted from the BIM model. Subsequently, through preprocessing such as signal filtering and noise reduction, data cleaning to remove outliers, timestamp alignment of multi-source data, and fusion standardization, a dataset that meets the requirements of subsequent processing is formed.

[0029] S2: Based on the original radio frequency signals and environmental data, construct and update the digital twin map of wireless signals in the construction environment in real time, and use the map to adaptively calibrate the tag signals to calculate the high-precision spatiotemporal trajectory of the target.

[0030] Preferably, the specific content of constructing and updating the digital twin map of wireless signals in the construction environment in real time is as follows: Obtain the training dataset, which includes the tag radio frequency signal features and their corresponding actual tag location coordinates collected under multiple sets of data from different environments.

[0031] It should be noted that the tag radio frequency signal characteristics refer to the physical layer parameters extracted from the tag's radio frequency signal that can be used to characterize its spatial location information, including: received signal strength indication, carrier phase, Doppler frequency shift, and signal arrival time difference.

[0032] It should be noted that the construction of the training dataset is the foundation for creating a high-precision digital twin map. In specific implementation, data collection needs to be carried out before deployment in the construction area or during closed construction periods. The data is read by a reader at known coordinate points by controlling the tags and recording each reading event.

[0033] Specifically, before construction begins, staff carry reference tags and move around the known coordinate points. Readers record the signal characteristics and environmental data at each location, and the reference coordinates are precisely measured using a total station with errors controlled within the centimeter level.

[0034] Furthermore, the data packet for each read event includes: Input feature vector: contains the RF signal characteristics of the tag at the time of the reading, such as received signal strength, carrier phase, and Doppler shift; as well as environmental data at that time, such as temperature, humidity, and electromagnetic interference intensity.

[0035] Output target value: the known three-dimensional spatial coordinates of the label.

[0036] By collecting a large number of data packets from different locations and under different environmental conditions, a training dataset is finally formed.

[0037] Based on machine learning algorithms, a positioning calibration model is trained using a training dataset. This model is used to characterize the mapping relationship between radio frequency signal characteristics, environmental data, and spatial location.

[0038] Furthermore, the positioning calibration model is built using a Gaussian process regression algorithm. The training process of this model involves learning a complex nonlinear mapping relationship from the input feature space (including radio frequency signal features and environmental data) to the output space (the actual location coordinates of the tag or signal propagation parameters). For new real-time input data, the calculation process of this model is as follows: First, the similarity between the new data and all training samples is calculated based on historical training data (using a preset kernel function, such as a radial basis function); then, the output values ​​of the training samples are weighted and averaged according to these similarities to obtain the predicted output value corresponding to the new data; finally, the model also calculates the uncertainty (variance) of this predicted value, which reflects the reliability of the prediction result. The final output of the model is the predicted value and its uncertainty estimate.

[0039] It should be noted that the on-site geometric structure (such as tunnel cross-sectional dimensions and supporting structures) and metal obstacle distribution information extracted from the construction BIM model are integrated as prior knowledge into the training process of the positioning calibration model. This information is used to define the physical constraints of signal propagation. For example, during the training of the positioning calibration model, for reader-tag pairs whose ray paths are blocked by walls or large metal components in the BIM model, their signal characteristics will be given lower weights or specially marked, so that the machine learning model can learn the signal propagation laws under complex spatial structures more accurately.

[0040] The real-time collected radio frequency signals and environmental data are input into the trained positioning calibration model to calculate the signal propagation parameters of the current environment.

[0041] It should be noted that when constructing the training dataset and the positioning calibration model, the on-site geometric structure and metal obstacle distribution information in the environmental data are used to determine the physical constraints of signal propagation, serving as important input features for the machine learning model to more accurately characterize the signal propagation characteristics at a specific spatial location.

[0042] Based on signal propagation parameters, a digital twin map is generated and updated to characterize the propagation characteristics of wireless signals throughout the construction area.

[0043] Furthermore, when changes in environmental data exceed a threshold, a local map update is automatically triggered, and a full map update is performed weekly to incorporate new training data and maintain the accuracy of the map.

[0044] It should be noted that individual and comprehensive threshold criteria for changes in environmental data (such as temperature change ±5) are established. 。 C. Electromagnetic interference changes of ±10dBm, or the Environmental Comprehensive Change Index (ECI) > 1.0, automatically trigger efficient local map updates and adopt a weekly global update mechanism. Utilizing verified high-precision positioning data automatically collected during system operation as new training samples, the positioning calibration model and digital twin map are reconstructed and optimized, thus forming a closed-loop self-optimizing system integrating real-time adaptive response and periodic iterative evolution. This effectively ensures the continuous accuracy and reliability of the digital twin map in complex and dynamic construction environments.

[0045] Preferably, the tag signal is adaptively calibrated using a digital twin map to calculate the target's high-precision spatiotemporal trajectory, specifically including the following steps: It can acquire the raw radio frequency signals of the tags and the current environmental data in real time within the construction area.

[0046] Based on the current environmental data, the signal propagation parameters of the corresponding area are queried from the digital twin map, including path loss coefficient, multipath attenuation factor and environmental interference weight.

[0047] It should be noted that during the real-time positioning phase, the system continuously collects the raw radio frequency signals emitted by the tags through the RFID reader network deployed in the construction area. Simultaneously, environmental monitoring sensors collect environmental data such as temperature, humidity, and electromagnetic interference intensity in real time. Based on this environmental data, the system queries the signal propagation parameters under the current environmental conditions from the pre-constructed digital twin map. These parameters include, but are not limited to: path loss coefficient (characterizing the signal attenuation characteristics with distance), multipath attenuation factor (reflecting the superposition effect of the signal after propagation through multiple paths), and environmental interference weight (quantifying the degree of influence of electromagnetic noise on the signal).

[0048] Based on the obtained signal propagation parameters, the original radio frequency signal is compensated and calibrated to eliminate the influence of environmental factors on signal measurement.

[0049] It should be noted that, based on the queried signal propagation parameters, the system performs real-time compensation calibration on the original radio frequency signal. First, it compensates for the received signal strength based on the path loss coefficient to eliminate signal attenuation caused by propagation distance. Then, it uses a multipath attenuation factor to correct signal phase distortion and compensate for measurement errors caused by multipath effects. Finally, it filters the signal based on environmental interference weights to suppress the influence of electromagnetic noise. Through these three calibration steps, the interference of environmental factors on signal measurement is effectively eliminated, resulting in more accurate signal characteristic values.

[0050] A weighted trilateration algorithm is used to calculate the preliminary spatial coordinates of the tag using the calibrated signal strength values.

[0051] It should be noted that the preliminary spatial coordinates of the tag are calculated using the calibrated signal strength value and an improved weighted trilateration algorithm. The specific calculation process is as follows: First, based on the calibrated signal strength value, the estimated distance from the tag to each reader is solved using a signal propagation model. The key parameter of the model, the path loss index, is obtained in real time from the digital twin map. Second, a weight is assigned to each reader based on the signal-to-noise ratio (SNR) of the received signal; the reader with the higher the SNR, the greater its weight. Finally, a weighted least squares method is used to solve an optimization problem. The goal of this problem is to find a spatial coordinate point that minimizes the sum of the weighted squared errors between the distance from this point to each reader and the aforementioned estimated distances. The resulting coordinates are the preliminary spatial coordinate estimates of the tag.

[0052] By combining data from the inertial measurement unit, the initial spatial coordinates are spatiotemporally fused using the Kalman filter algorithm to generate a smooth, continuous, and high-precision spatiotemporal trajectory.

[0053] It should be noted that the inertial measurement unit data is collected by the IMU sensors in the smart safety helmet, wristband and mechanical vehicle terminal worn by the personnel, including motion information such as three-axis acceleration and three-axis angular velocity, and is transmitted to the central processing server through wireless communication protocols (such as ZigBee, LoRa or 5G) for timestamp alignment and fusion with radio frequency signals and environmental data.

[0054] It should be noted that, in order to obtain a smoother and more accurate motion trajectory, the system further integrates inertial measurement unit (IMU) data. Through the Kalman filter algorithm, the spatial coordinates obtained by RFID positioning are fused with the acceleration and angular velocity information provided by the IMU. The state vector of the Kalman filter includes position, velocity, and acceleration, and the measurement vector is the RFID positioning coordinates and IMU data. Through the iterative process of prediction and correction, the random error of RFID positioning is effectively suppressed, the cumulative error of the IMU is compensated, and finally high-precision, high-refresh-rate spatiotemporal trajectory data is output.

[0055] It should be noted that the generated spatiotemporal trajectory needs to be post-processed to further improve its quality. Sliding window mean filtering is used to eliminate high-frequency noise in the trajectory, motion constraint algorithm is used to ensure that the trajectory conforms to the laws of physical motion, and finally, interpolation algorithm is used to ensure the uniformity and continuity of the trajectory time series. These processes make the final spatiotemporal trajectory both accurate and smooth, providing a reliable data foundation for subsequent safety risk assessment.

[0056] For example, before construction, technicians carry reference RFID tags and move around 100 centimeter-precision coordinate points calibrated by a total station in the tunnel to collect multiple sets of radio frequency signal characteristics and environmental data as training data. The Gaussian process regression algorithm is used to train the positioning calibration model. During construction, real-time radio frequency signals and environmental data are input into the model to obtain signal propagation parameters (path loss coefficient, multipath attenuation factor, etc.), generating and dynamically updating a digital twin map of wireless signals across the entire tunnel (local updates are triggered when environmental data exceeds a threshold, and the entire map is updated weekly). The original radio frequency signals are then calibrated based on the map parameters, and preliminary coordinates are calculated using an improved weighted trilateration algorithm. The IMU data is then fused, Kalman filtered, and processed in a spatiotemporal manner to generate a smooth, continuous, high-precision spatiotemporal trajectory (error < 3cm) for targets such as excavators.

[0057] S3: Input high-precision spatiotemporal trajectories into the spatiotemporal relationship inference engine, perform inferences based on the built-in physical rule model, and output a quantified safety risk level.

[0058] Preferably, the specific process of the spatiotemporal relationship reasoning engine in deduction includes: A spatiotemporal relationship matrix of human-machine environment is established to quantify the spatiotemporal correlation between personnel, machinery, and environmental elements. The calculation of the spatiotemporal correlation adopts a composite function based on distance and motion state, and the specific formula is as follows: ; In the formula, For object and At any moment The spatiotemporal correlation quantifies the degree of proximity and movement trend of two targets in space and time; the higher the value, the greater the risk. The weighting coefficients for the distance term are determined through optimization using historical data, adjusting the importance of the distance factor in the overall correlation. For object and The Euclidean distance between the two targets is obtained in real time from the RFID positioning system. Calculating the absolute spatial distance between the two targets is a core element in assessing collision risk. This is a distance adjustment parameter, set according to safety distance requirements, controlling the decay rate of the distance item, and determining at what distance it begins to have a significant impact on risk; The weighting coefficient for the speed term is determined through optimization using historical data, adjusting the importance of the relative speed factor in the overall correlation. For object and The relative velocity between them is calculated by differential analysis of the IMU and the positioning trajectory. The greater the velocity difference, the higher the severity of the potential collision and the greater the risk. The speed adjustment parameter is set according to typical motion speeds, controlling the decay rate of the speed term and determining the magnitude of the speed difference that begins to have a significant impact on risk. The weighting parameters for the direction term are determined through optimization using historical data, adjusting the importance of motion direction factors in the overall correlation. The angle between the moving directions of the object and the target is calculated from the IMU and the positioning trajectory. The larger the angle, the more consistent the moving directions are, and the higher the probability of chasing or colliding.

[0059] The probability of potential conflict is calculated based on spatiotemporal correlation, using the following formula: ; In the formula, For object and Between moments The probability of potential conflict; The conflict sensitivity coefficient is set based on historical accident data or expert experience. For object and At any moment The spatiotemporal correlation quantifies the degree of proximity and movement trend of two targets in space and time; the higher the value, the greater the risk. For object and The Euclidean distance between the two targets is obtained in real time from the RFID positioning system. Calculating the absolute spatial distance between the two targets is a core element in assessing collision risk. The value should be a very small positive number to prevent the denominator from being zero. For object and The relative velocity between them is calculated by differential analysis of the IMU and the positioning trajectory. The greater the velocity difference, the higher the severity of the potential collision and the greater the risk. The maximum speed threshold is preset by the system and set according to the type of construction area, the content of the work, and safety regulations.

[0060] A multi-level risk assessment algorithm is adopted to generate a quantitative security risk level by combining spatiotemporal correlation and conflict probability.

[0061] Specifically, the details of the security risk levels are as follows: The high-risk level is determined by the probability of potential conflict. Greater than or equal to the high-risk probability threshold or the Euclidean distance between objects If the distance is less than or equal to the high-risk distance threshold, and any one of these conditions is met, it is considered high-risk, indicating that a collision or safety accident is likely to occur, and immediate emergency intervention measures are required.

[0062] The medium-risk level is determined by the following criteria: it does not reach the high-risk level and meets the potential conflict probability. Greater than or equal to the medium-risk probability threshold and less than the high-risk probability threshold (the medium-risk probability threshold is less than the high-risk probability threshold), or the Euclidean distance between objects. If the distance threshold is less than or equal to the medium-risk distance threshold (which is greater than the high-risk distance threshold), this level indicates a potential dangerous trend, and an early warning should be issued to alert relevant personnel.

[0063] The criteria for determining a low-risk level are: not reaching the high-risk or medium-risk levels, i.e., the probability of potential conflict is low. Less than the medium risk probability threshold and the Euclidean distance between objects If the distance is greater than the medium-risk distance threshold, this level indicates that the spatiotemporal interaction between targets is weak and they are in a normal operating state. The system only needs to keep monitoring them.

[0064] Furthermore, construction schedule data is used to drive the dynamic updates of digital twin maps and risk rules. The system predicts construction activities in a certain area at a certain time in the future (such as 'large-scale hoisting operations') based on the schedule, thereby calling the corresponding signal propagation model library and stricter risk threshold parameters in advance to achieve forward-looking adjustments to risk warnings.

[0065] It should be noted that the specific details of the emergency intervention measures are as follows: Level 1 Intervention (Audio-Visual Alarm and Terminal Prompt): The system immediately sends instructions to the on-site broadcast system and intelligent warning lights in the core risk area, triggering a high-decibel alarm and flashing red lights to warn all personnel and machine operators in the area. At the same time, it sends the highest priority alarm information to the smart safety helmets, handheld terminals, and vehicle-mounted displays in the machine cabs of relevant personnel. The information includes the type of danger (e.g., "collision warning"), the dangerous object (e.g., "approaching excavator"), direction, and distance.

[0066] Level 2 Intervention (Mandatory Speed ​​Limit and Braking): For intelligent construction machinery (such as excavators and pavers) equipped with onboard control systems, the system sends mandatory commands to the machinery's controller via an IoT interface. If the risk level is high and continues to escalate, the system first executes mandatory speed limiting to restrict the machinery's power output within a safe range. If the probability of conflict is extremely high and cannot be avoided, an emergency braking command is sent to stop the machinery from operating until the risk is eliminated.

[0067] Level 3 intervention (linked environmental control): The system links with other intelligent subsystems in the construction area. For example, it automatically sends avoidance instructions to tower crane operators in relevant areas and restricts their operating radius; or it sends "stop operation" instructions to construction elevators to prevent the risk from expanding in densely populated areas.

[0068] Furthermore, the high-risk distance threshold is obtained by taking into account the minimum safe working distance clearly defined in the Technical Regulations for Safety of Railway Engineering Construction, and by combining it with the technical parameters of specific construction machinery (such as the maximum working radius). During operation, this threshold will be dynamically adjusted according to the real-time construction activity type and environmental visibility data.

[0069] The medium-risk distance threshold is determined by adding a dynamic buffer distance to the high-risk distance threshold. The logical parameters of this buffer distance (such as a multiple or a fixed value) are matched according to the complexity of the construction area (such as the population density), and its value changes in conjunction with the dynamic adjustment of the high-risk distance threshold.

[0070] The high-risk probability threshold is obtained by statistical analysis of the historical safety event dataset pre-stored in the system. Specifically, it uses the high-order quantile (such as the 95th quantile) of the conflict probability values ​​corresponding to all real dangerous events in the dataset. After the system is running, this threshold will be adaptively optimized based on the operational risk level and the accuracy feedback of historical warnings.

[0071] The medium-risk probability threshold is obtained by calculating the high-risk probability threshold using a preset fixed ratio coefficient to ensure the hierarchical nature of risk level determination. This ratio coefficient is a constant set based on engineering experience. In practical applications, this threshold is dynamically adjusted in conjunction with the high-risk probability threshold.

[0072] Establish a dynamic risk threshold adjustment mechanism to adaptively adjust risk assessment standards based on construction progress and environmental changes.

[0073] Specifically, the details of the dynamic risk threshold adjustment mechanism are as follows: When the system identifies the current operation as a high-risk process (such as "large component hoisting") or the environmental conditions as harsh (such as visibility below the set value or electromagnetic interference exceeding the set limit), it automatically performs strict adjustments: that is, it lowers the probability thresholds for high and medium risks, while raising the distance thresholds for high and medium risks, making the system more sensitive to judgment and providing earlier warnings.

[0074] When the system is in a low-risk routine operation and the environmental conditions are good, it automatically performs a relaxation adjustment: that is, it raises the probability threshold and lowers the distance threshold to reduce the false alarm rate and reduce interference with normal operations while ensuring safety.

[0075] It should be noted that one of the bases for the dynamic risk threshold adjustment mechanism is the construction schedule extracted from the BIM model. When the schedule shows that the current process has entered a 'high-risk process' (such as the 'large component hoisting' node in the BIM plan), the system automatically triggers the threshold adjustment process and executes a strict adjustment strategy.

[0076] For example, the high-precision spatiotemporal trajectories (including position, speed, and direction of movement) of the worker and the excavator inside the tunnel are input into the spatiotemporal relationship inference engine. The spatiotemporal correlation formula with preset weight coefficients is substituted to calculate the spatiotemporal correlation corresponding to the Euclidean distance, relative speed, and directional angle between the two. Then, based on the correlation and preset parameters (conflict sensitivity coefficient, maximum speed threshold, etc.), the probability of potential conflict is calculated. Combined with dynamic risk thresholds (high risk probability threshold 10%, distance threshold 2 meters, medium risk probability threshold 3%, distance threshold 3 meters), it is determined that the worker and the excavator are in a medium risk state. Then, an audible and visual warning is sent to the worker's smart safety helmet and a deceleration prompt is pushed to the excavator's cab.

[0077] S4: Based on the feedback results of risk warning, the digital twin map and physical rule model are synergistically iteratively optimized to realize the self-evolution of the dynamic prediction method for railway construction safety risks.

[0078] Preferably, the collaborative iterative optimization process employs an adaptive optimization framework based on reinforcement learning, specifically including: An optimization objective function is established, comprehensively considering multiple performance indicators such as positioning accuracy, early warning accuracy, and system response time. The specific formula is as follows: ; In the formula, For the overall performance indicators of the system; As a balancing coefficient, it is used to balance the contribution weights of the three types of indicators, namely "positioning accuracy, early warning accuracy, and system response time", in the objective function, so as to avoid a single indicator from excessively dominating the optimization direction; For the integration operator and the integration field, The integration domain is the range of values ​​for the parameter to be optimized. As an integral variable, it represents the set of all optimizable parameters of the "digital twin map + physical rule model", realizing global parameter optimization and avoiding optimization being limited to local parameters; The positioning error vector is used to directly quantify the positioning accuracy. This provides a quantifiable standard for positioning accuracy, ensuring that the positioning error can be directly substituted into the function for optimization. The early warning accuracy rate is a metric that is calculated in real time to quantify the early warning accuracy rate. The weighting function for early warning accuracy is set based on historical data fitting or engineering experience to avoid the early warning accuracy being "marginalized" in the optimization of multiple indicators; The system response time metric is used when the model parameters are... The total time taken by the system from "RFID reader collecting tag signal" to "terminal outputting risk level" reflects the system's real-time performance, and the system log records this in real time. The response time penalty function is set based on the safety response requirements of railway construction to prevent the system from sacrificing response time in order to improve positioning accuracy / early warning accuracy.

[0079] Design a reward mechanism to dynamically adjust the update frequency of the digital twin map and the parameters of the physical rule model based on the effectiveness of risk warnings. The specific formula is as follows: ; In the formula, for The reward value at any given moment represents the optimization effect of the current parameter combination. The better the optimization effect, the worse the effect. It serves as a target guidance signal for collaborative iterative optimization. As a true example rate weighting coefficient, based on the safety priority preset of railway construction scenarios, the positive contribution of "correct early warning" to the reward value is amplified, guiding the system to prioritize ensuring the accuracy of early warnings; for Real-time rate; The false positive rate weighting coefficient is based on the efficiency requirements preset in the railway construction scenario. for The rate of false positives at any given time; The response time increment weighting coefficient is preset based on the handling requirements of emergency risks in railway construction; for Increment of system response time at any moment; The weighting coefficients for the underreporting indicator function are preset based on the severe consequences of railway construction safety accidents; for Time-lapse indicator function.

[0080] A distributed optimization strategy is adopted, which performs local optimization on edge computing nodes and global model aggregation in the cloud.

[0081] Specifically, edge computing nodes are deployed according to the distribution density of RFID readers within the construction area, with each edge computing node corresponding to a fixed number of RFID readers. In tunnel construction scenarios, additional edge computing nodes are added according to the length of the construction section to ensure no coverage blind spots. Each edge computing node is only responsible for optimization tasks within its jurisdiction, and the optimization targets are limited to two categories: Only the signal propagation parameters within the jurisdiction are updated, including path loss coefficient and multipath attenuation factor; the update of the entire construction area map is not involved.

[0082] The risk assessment threshold is adjusted only for the types of work currently being carried out within the jurisdiction (such as tunnel excavation, roadbed paving, and bridge hoisting), and the threshold parameters for other construction areas are not adjusted.

[0083] Specifically, edge computing nodes collect raw RFID radio frequency signals, environmental data (temperature, humidity, electromagnetic interference intensity), positioning error records, and risk warning logs within a set period in real time within their jurisdiction.

[0084] When the positioning error exceeds the preset accuracy requirement for multiple consecutive data collections, or the false alarm rate exceeds the preset control range, or the number of missed alarms reaches the preset trigger value, the local optimization process is initiated.

[0085] If the positioning error exceeds the limit when calling the lightweight gradient descent algorithm, adjust the signal compensation coefficient of the corresponding area in the digital twin map; if the false alarm rate exceeds the limit, adjust the medium-risk probability threshold; if the number of missed alarms exceeds the limit, adjust the high-risk probability threshold.

[0086] Record the positioning error value, false alarm rate, and number of missed alarms before and after optimization, generate the "Local Optimization Result Table", and store it in the local database of the edge computing node.

[0087] Furthermore, during construction breaks (such as fixed time periods each day), all edge computing nodes upload the "Local Optimization Result Table" and the adjusted parameters to the cloud platform.

[0088] The cloud platform assigns weights based on the risk level of the areas under the jurisdiction of edge computing nodes. The node parameter weights are higher in high-risk areas (such as tunnel blasting areas and bridge hoisting areas) than in medium-risk areas (such as roadbed paving areas), and higher in medium-risk areas than in low-risk areas (such as material storage areas).

[0089] By employing a federated learning algorithm, the adjusted parameters of each edge node are weighted and averaged according to the assigned weights to obtain unified digital twin map parameters and physical rule model parameters for the entire construction area.

[0090] If the adjusted parameters of an edge node deviate from the global weighted average parameters beyond the preset range, the cloud platform will trigger a manual review process. If the review is successful, the node's parameters will be included in the global parameters; if the review fails, the node's parameters will be rejected and the local optimization will be re-executed.

[0091] The cloud platform synchronizes the calculated global parameters to all edge computing nodes. Upon receiving the parameters, the edge computing nodes immediately replace the local parameters to ensure the consistency of parameters across the entire construction area.

[0092] Establish a model performance evaluation system and regularly evaluate and adjust the optimization results.

[0093] Specifically, after the daily construction ends, each edge computing node automatically calculates the above five indicators within its jurisdiction during a fixed period, generates a "Daily Performance Self-Assessment Report", and uploads it to the cloud platform.

[0094] During a fixed time each week, the cloud platform compiles the "Daily Performance Self-Assessment Report" of all edge computing nodes, calculates the average indicators of the entire construction area, and generates a "Weekly Global Assessment Report". The assessment results are divided into three categories: "Excellent", "Qualified", and "Unqualified". "Excellent" means that all indicators meet the qualified standard; "Qualified" means that only one indicator does not meet the qualified standard; and "Unqualified" means that two or more indicators do not meet the qualified standard.

[0095] Furthermore, the excellent results feedback adjustment: maintain the current digital twin map update frequency, edge node local optimization cycle, and physical rule model parameter optimization frequency.

[0096] Feedback and adjustment of qualified results: Targeted optimization for indicators that do not meet the qualified standards. If the accuracy of the early warning is not up to standard, the weight coefficient of the true case rate is adjusted on the cloud platform; if the system response time is not up to standard, the edge computing nodes perform lightweight processing on the model.

[0097] Feedback and adjustment for unqualified results: Trigger the "emergency optimization mode" to shorten the update frequency of the entire digital twin map and the local optimization cycle of edge nodes. At the same time, supplement the risk event data within the set period to retrain the physical rule model. After the retraining is completed, a special evaluation will be carried out the next day until the evaluation result reaches the "qualified" standard or above.

[0098] All Daily Performance Self-Assessment Reports, Weekly Global Assessment Reports, and corresponding adjustment measures are stored in the system log. The log retention period meets the requirements for traceability of railway construction data and is used for reference in subsequent optimization work.

[0099] For example, a local optimization was initiated at an edge node in the middle section of the tunnel (managing 3 readers) due to a positioning error exceeding 5cm and 2 false alarms. Relevant data was collected within 1 hour, and the path loss coefficient and medium-risk probability threshold of the area were adjusted using a lightweight gradient descent algorithm. After optimization, the positioning error was reduced to 4cm, false alarms were cleared, and the results were recorded. During daily construction breaks, all edge nodes uploaded the optimization results to the cloud. The cloud assigned weights according to the regional risk level (e.g., 0.3 for the bridge hoisting area), and obtained unified parameters for the entire region through federated learning weighted averaging (manual review was triggered if the parameter deviation exceeded 20%), which were then synchronized to all edge nodes. Weekly summaries and evaluations were conducted. If the overall indicators were excellent, the current optimization cycle was maintained. If the early warning accuracy was not up to standard, emergency optimization was triggered (shortening the map update frequency and retraining the model) until the evaluation was satisfactory.

[0100] In summary, this invention achieves precise calibration and compensation of RFID signals in complex construction environments by constructing a dynamically updated digital twin map of wireless signals, significantly improving the accuracy and stability of spatiotemporal trajectory positioning for personnel and machinery. It utilizes a spatiotemporal relationship inference engine to fuse multidimensional data for forward-looking risk simulation, outputting quantified safety risk levels and realizing a shift from passive alarm to proactive early warning. In particular, the adoption of a reinforcement learning-based collaborative iterative optimization mechanism enables the system to continuously evolve based on early warning feedback, continuously improving positioning accuracy, early warning accuracy, and response speed, ultimately forming a highly reliable, self-learning, and adaptive high-precision proactive protection system for railway construction safety.

[0101] Example 2, an embodiment of the present invention, provides an intelligent management system for railway construction personnel and machinery information based on RFID, comprising: a data acquisition module for collecting raw radio frequency signals and environmental data of personnel and machinery tags within the construction area via a radio frequency identification reader network; a map construction module for constructing and updating a digital twin map of wireless signals in the construction environment in real time based on the raw radio frequency signals and environmental data, and using the map to adaptively calibrate the tag signals to calculate the high-precision spatiotemporal trajectory of the target; a level output module for inputting the high-precision spatiotemporal trajectory into a spatiotemporal relationship inference engine, performing deduction based on a built-in physical rule model, and outputting a quantified safety risk level; and an iterative optimization module for performing collaborative iterative optimization of the digital twin map and the physical rule model based on the feedback results of risk warnings, thereby realizing intelligent management of railway construction personnel and machinery information.

[0102] Example 3 is an embodiment of the present invention, which differs from the previous embodiment in that: If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0103] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0104] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0105] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0106] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for intelligent management of human-machine information in railway construction based on RFID, characterized in that: include: The original radio frequency signals and environmental data of personnel and machinery tags in the construction area are collected through a network of radio frequency identification readers; Based on the original radio frequency signals and environmental data, a digital twin map of wireless signals in the construction environment is constructed and updated in real time. The map is then used to adaptively calibrate the tag signals and calculate the high-precision spatiotemporal trajectory of the target. The high-precision spatiotemporal trajectory is input into the spatiotemporal relationship reasoning engine, which performs deductions based on the built-in physical rule model and outputs a quantified safety risk level. Based on the feedback results of risk warning, the digital twin map and physical rule model are collaboratively and iteratively optimized to realize intelligent management of human and machine information in railway construction. The environmental data includes temperature, humidity, noise intensity, electromagnetic interference level, satellite positioning signal quality, as well as on-site geometric structure, metal obstacle distribution information, and construction schedule data extracted from the construction BIM model. The specific details of constructing and updating the digital twin map of wireless signals in the construction environment in real time are as follows: Obtain a training dataset, which includes the tag radio frequency signal features collected under multiple sets of different environmental data and their corresponding actual tag location coordinates; Based on machine learning algorithms, a positioning calibration model is trained using the training dataset. The positioning calibration model is used to characterize the mapping relationship between radio frequency signal characteristics, environmental data and spatial location. The real-time collected radio frequency signals and environmental data are input into the trained positioning calibration model to calculate the signal propagation parameters of the current environment; Based on the signal propagation parameters, the digital twin map used to characterize the wireless signal propagation characteristics of the entire construction area is generated and updated; The process of adaptively calibrating the tag signal using a digital twin map to calculate the target's high-precision spatiotemporal trajectory includes the following steps: Real-time acquisition of raw radio frequency signals and current environmental data from tags within the construction area; Based on the current environmental data, query the signal propagation parameters of the corresponding area from the digital twin map, including path loss coefficient, multipath attenuation factor and environmental interference weight; Based on the obtained signal propagation parameters, the original radio frequency signal is compensated and calibrated to eliminate the influence of environmental factors on signal measurement; a weighted trilateration algorithm is used to calculate the preliminary spatial coordinates of the tag using the calibrated signal strength value; By combining data from the inertial measurement unit, the initial spatial coordinates are spatiotemporally fused using the Kalman filter algorithm to generate a smooth, continuous, and high-precision spatiotemporal trajectory.

2. The intelligent management method for railway construction human-machine information based on RFID as described in claim 1, characterized in that: The specific process of the spatiotemporal relationship reasoning engine includes: Establish a spatiotemporal relationship matrix between humans, machines, and the environment to quantify the spatiotemporal correlation between personnel, machinery, and environmental elements; Calculate the probability of potential conflicts based on spatiotemporal correlation. A multi-level risk assessment algorithm is adopted to generate a quantitative security risk level by combining spatiotemporal correlation and conflict probability. Establish a dynamic risk threshold adjustment mechanism to adaptively adjust risk assessment standards based on construction progress and environmental changes.

3. The intelligent management method for railway construction human-machine information based on RFID as described in claim 2, characterized in that: The specific details of the security risk level are as follows: The high-risk level is determined by the following conditions: the probability of potential conflict is greater than or equal to the high-risk probability threshold or the Euclidean distance between objects is less than or equal to the high-risk distance threshold. If either condition is met, it is determined to be high-risk, which indicates that a collision or safety accident will occur and emergency intervention measures must be taken immediately. The medium risk level is determined by the following conditions: it does not reach the high risk level, and the potential conflict probability is greater than or equal to the medium risk probability threshold and less than the high risk probability threshold, or the Euclidean distance between objects is less than or equal to the medium risk distance threshold. This level indicates that there is a potential dangerous trend, and an early warning should be issued to remind relevant personnel to pay attention. The low-risk level is determined by the following criteria: it does not reach the high-risk or medium-risk levels, that is, the potential conflict probability is less than the medium-risk probability threshold and the Euclidean distance between objects is greater than the medium-risk distance threshold. This level indicates that the spatiotemporal interaction between targets is weak and they are in a normal operating state. The system only needs to keep monitoring them.

4. The intelligent management method for railway construction human-machine information based on RFID as described in claim 3, characterized in that: The collaborative iterative optimization process employs an adaptive optimization framework based on reinforcement learning, specifically including: Establish an optimization objective function that comprehensively considers multiple performance indicators such as positioning accuracy, early warning accuracy, and system response time. Design a reward mechanism to dynamically adjust the update frequency of the digital twin graph and the parameters of the physical rule model based on the effect of risk warning; adopt a distributed optimization strategy to perform local optimization on edge computing nodes and global model aggregation in the cloud; Establish a model performance evaluation system and regularly evaluate and adjust the optimization results.

5. An RFID-based intelligent management system for railway construction human-machine information, based on the RFID-based intelligent management method for railway construction human-machine information according to any one of claims 1 to 4, characterized in that: include, The data acquisition module is used to collect raw radio frequency signals from personnel and machinery tags within the construction area via a network of radio frequency identification (RFID) readers. The system includes a tag signal and environmental data module; a map construction module, which is used to construct and update a digital twin map of wireless signals in the construction environment in real time based on the original radio frequency signals and environmental data, and to use the map to adaptively calibrate the tag signals and calculate the high-precision spatiotemporal trajectory of the target. The level output module is used to input high-precision spatiotemporal trajectories into the spatiotemporal relationship inference engine, perform inferences based on the built-in physical rule model, and output a quantified safety risk level. The iterative optimization module is used to perform collaborative iterative optimization of the digital twin map and the physical rule model based on the feedback results of risk warning, so as to realize intelligent management of human and machine information in railway construction.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the RFID-based intelligent management method for railway construction human-machine information as described in any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the RFID-based intelligent management method for railway construction human-machine information as described in any one of claims 1 to 4.

Images