A real-time monitoring system for safety posture of railway workers

Abstract

The application relates to a railway worker safety posture real-time monitoring system, method, device, computer equipment, readable storage medium and program product, and relates to the technical field of railway safety monitoring. The application can improve discrimination accuracy and reduce false positive rate. The system comprises a video access analysis module, a railway operation posture recognition module, a business fusion linkage module and a pre-warning closed-loop management module. The video access analysis module is used for performing inference analysis on workers by using a light-weight posture recognition inference unit deployed on an edge side and video streams obtained through video monitoring resources. The railway operation posture recognition module is used for centrally managing posture monitoring data for railway operation posts. The business fusion linkage module is used for associating and matching posture recognition results generated by the posture recognition unit with multi-dimensional context data of railway operations. The pre-warning closed-loop management module is used for generating a graded pre-warning event for behaviors determined as unsafe postures of railway operations according to safety control rules, pushing the event to corresponding monitoring post terminals, and receiving and recording obtained disposal feedback information.

Description

Technical Field

[0001] This application relates to the field of railway safety monitoring technology, and in particular to a real-time monitoring system, method, device, computer equipment, computer-readable storage medium, and computer program product for the safe posture of railway workers. Background Technology

[0002] With the continuous expansion of railway transportation networks and the increasing trend towards heavy-haul and high-frequency operations, railway construction and daily maintenance scenarios are becoming increasingly complex. The large number, wide distribution, and high mobility of on-site workers have significantly increased the pressure on safety management. Simultaneously, the requirements for safe production are constantly rising, driving the railway industry to accelerate its transformation towards intelligence and digitalization. Against this backdrop, how to utilize video intelligent analysis methods to monitor the behavior and posture of railway workers in real time, accurately, and automatically has become a key requirement for ensuring railway operational safety and preventing personal injury accidents.

[0003] However, existing video analytics technologies are mostly focused on general security scenarios, such as intrusion detection and loitering, and are difficult to adapt to the specific operational procedures and safety standards of railways. In particular, they suffer from low accuracy and high false alarm rates in identifying key safety postures such as bending over to work, running across tracks, and improper standing postures of safety personnel, and cannot achieve timely early warning and closed-loop handling of unsafe posture behaviors. Summary of the Invention

[0004] Based on this, it is necessary to provide a real-time monitoring system, method, device, computer equipment, computer-readable storage medium, and computer program product for the safe posture of railway workers, addressing the aforementioned technical problems.

[0005] In a first aspect, this application provides a real-time monitoring system for the safe posture of railway workers, the system comprising: a video access and analysis module, a railway work posture recognition module, a business integration and linkage module, and an early warning closed-loop management module; wherein...

[0006] The video access and analysis module is used to access video surveillance resources along the railway line and use a lightweight attitude recognition inference unit deployed on the edge side to perform inference analysis on the workers through the video stream obtained from the video surveillance resources.

[0007] The railway operation posture recognition module is used to centrally manage posture monitoring data for railway operation positions, and to train posture recognition units in railway scenarios based on the posture monitoring data.

[0008] The business integration and linkage module is used to associate and match the posture recognition results generated by the posture recognition unit with the multi-dimensional context data of railway operations, and automatically determine whether the currently identified unsafe posture behavior of railway operations is within the scope of railway operation compliance context constraints.

[0009] The early warning closed-loop management module is used to generate graded early warning events for behaviors judged as unsafe postures in railway operations according to safety control rules, push them to the corresponding monitoring terminal, receive and record the obtained handling feedback information, and form a closed-loop management mechanism for the entire process.

[0010] In one embodiment, the video access analysis module includes an access adaptation submodule, a containerized operation submodule, and a posture inference execution submodule. The access adaptation submodule is used to be compatible with video surveillance resources along the railway line and uniformly convert them into standardized video streams. The containerized operation submodule is used to build an edge containerized environment based on integrated security management at the edge. The posture inference execution submodule is used to call the deployed lightweight posture recognition inference unit in the edge containerized environment based on the standardized video stream to perform real-time analysis of the spatiotemporal characteristics of the human posture of the workers in the standardized video stream and output the identification results of unsafe posture behaviors in railway operations.

[0011] In one embodiment, the railway operation posture recognition module includes a behavior standard definition submodule, an intelligent training interaction submodule, and a version management submodule. The behavior standard definition submodule is used to define the categories and discrimination rules of unsafe posture behaviors for posture monitoring of railway operation positions, providing a standardized supervision basis for the training of posture recognition intelligent analysis units. The intelligent training interaction submodule is used to provide a user interface, supporting on-site maintenance personnel to perform intelligent training task arrangement on target posture behaviors based on collected railway operation video clips without programming capabilities, and calling the underlying training engine to generate or optimize the posture recognition intelligent analysis units suitable for line conditions. The version management submodule is used to perform intelligent analysis and full-cycle governance of multiple posture recognition intelligent analysis units trained in the railway scenario, and through linkage with the containerized operation submodule, complete the push update of the specified version analysis unit to the edge-side integrated security management.

[0012] In one embodiment, the business integration and linkage module includes: a context data access submodule, an association rule engine submodule, and a compliance judgment submodule; the context data access submodule is used to connect to the multi-source business of railway operations and the spatial information system, collect and process the four-tuple of railway operation compliance context, and form a unified format of railway operation multi-dimensional context data stream; the association rule engine submodule is used to build a logical rule library, map and bind the types of unsafe posture behaviors of railway operations with the corresponding compliance context constraints, and dynamically adjust the rule parameters through the management interface; the compliance judgment submodule is used to synchronously obtain the current railway operation multi-dimensional context data after receiving the recognition result output by the posture reasoning execution submodule, perform matching logic judgment based on the mapping relationship in the posture context association rule engine, output a binary judgment result of whether the target behavior is within the scope of compliance context constraints, and transmit the judgment result together with the railway operation multi-dimensional context data to the early warning closed-loop management module.

[0013] In one embodiment, the early warning closed-loop management module includes: a strategy configuration submodule, an alarm push submodule, and a handling tracking submodule;

[0014] The strategy configuration submodule is used to define three-level linkage response strategy tuples corresponding to different combinations of unsafe posture behaviors and compliance context judgment results in railway safety management rules. It dynamically adjusts the early warning and handling strategy templates through a visual interface to form an executable hierarchical early warning strategy library. The alarm push submodule is used to push the current early warning event to the monitoring terminal through a message middleware after receiving the violation judgment result output by the compliance judgment submodule and matching the corresponding early warning responsibility mapping pair according to the hierarchical early warning strategy library. The handling tracking submodule is used to record the full-link operation log of each early warning event, periodically summarize unclosed events and trigger the overdue supervision mechanism.

[0015] In one embodiment, the system further includes a cloud-edge collaborative strategy distribution module; the cloud-edge collaborative strategy distribution module is used to package the latest attitude recognition intelligent analysis unit and early warning and handling strategy template generated by the version management submodule and distribute them to the corresponding edge-side integrated security management and control, so as to complete the update of the attitude recognition intelligent analysis unit and the three-level linkage response strategy tuple, and synchronously transmit them back to the railway operation attitude recognition module.

[0016] Secondly, this application also provides a method for real-time monitoring of the safe posture of railway workers, applied to the real-time monitoring system for the safe posture of railway workers as described in the above embodiments, the method comprising:

[0017] The video access and analysis module accesses video surveillance resources along the railway line, and uses a lightweight attitude recognition inference unit deployed at the edge to perform inference analysis on the video stream obtained from the video surveillance resources for the workers.

[0018] The posture monitoring data for railway operation positions is centrally managed through the railway operation posture recognition module, and posture recognition units are trained and formed in railway scenarios based on the posture monitoring data.

[0019] The business integration and linkage module is used to associate and match the posture recognition results generated by the posture recognition unit with the multi-dimensional context data of railway operations, and automatically determine whether the currently identified unsafe posture behavior of railway operations is within the scope of railway operation compliance context constraints.

[0020] The early warning closed-loop management module generates graded early warning events based on safety control rules for behaviors deemed as unsafe postures in railway operations. These events are then pushed to the corresponding monitoring terminal, where the received feedback information is received and recorded, forming a closed-loop management mechanism for the entire process.

[0021] Thirdly, this application also provides a real-time monitoring device for the safe posture of railway workers, applied to the real-time monitoring system for the safe posture of railway workers as described in the above embodiments, the device comprising:

[0022] The reasoning and analysis unit is used to access video surveillance resources along the railway line through the video access and analysis module, and to perform reasoning and analysis on the workers by using the video stream obtained from the video surveillance resources through the lightweight attitude recognition reasoning unit deployed on the edge side.

[0023] The scenario training unit is used to centrally manage the posture monitoring data for railway operation positions through the railway operation posture recognition module, and to form a posture recognition unit in the railway scenario based on the posture monitoring data.

[0024] The association matching unit is used to use the business integration and linkage module to associate and match the posture recognition results generated by the posture recognition unit with the multi-dimensional context data of railway operations, and automatically determine whether the currently identified unsafe posture behavior of railway operations is within the scope of railway operation compliance context constraints.

[0025] The graded early warning unit is used by the early warning closed-loop management module to generate graded early warning events for behaviors that are judged to be unsafe postures in railway operations according to safety control rules, and push them to the corresponding monitoring terminal. The unit receives and records the feedback information of the handling, forming a closed-loop management mechanism for the entire process.

[0026] Fourthly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0027] The system accesses video surveillance resources along the railway line via a video access and analysis module. A lightweight attitude recognition inference unit deployed at the edge performs inference analysis on the video streams obtained from these resources. A railway operation attitude recognition module centrally manages attitude monitoring data for railway work positions. Based on this data, attitude recognition units are trained in railway scenarios. A business fusion and linkage module correlates and matches the attitude recognition results generated by the units with multi-dimensional contextual data of railway operations, automatically determining whether currently identified unsafe attitude behaviors occur within the compliance context constraints of railway operations. An early warning closed-loop management module generates tiered early warning events based on safety control rules for behaviors deemed unsafe, pushing them to the corresponding monitoring terminal. The module receives and records the feedback information, forming a closed-loop management mechanism throughout the entire process.

[0028] Fifthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0029] The system accesses video surveillance resources along the railway line via a video access and analysis module. A lightweight attitude recognition inference unit deployed at the edge performs inference analysis on the video streams obtained from these resources. A railway operation attitude recognition module centrally manages attitude monitoring data for railway work positions. Based on this data, attitude recognition units are trained in railway scenarios. A business fusion and linkage module correlates and matches the attitude recognition results generated by the units with multi-dimensional contextual data of railway operations, automatically determining whether currently identified unsafe attitude behaviors occur within the compliance context constraints of railway operations. An early warning closed-loop management module generates tiered early warning events based on safety control rules for behaviors deemed unsafe, pushing them to the corresponding monitoring terminal. The module receives and records the feedback information, forming a closed-loop management mechanism throughout the entire process.

[0030] Sixthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0031] The system accesses video surveillance resources along the railway line via a video access and analysis module. A lightweight attitude recognition inference unit deployed at the edge performs inference analysis on the video streams obtained from these resources. A railway operation attitude recognition module centrally manages attitude monitoring data for railway work positions. Based on this data, attitude recognition units are trained in railway scenarios. A business fusion and linkage module correlates and matches the attitude recognition results generated by the units with multi-dimensional contextual data of railway operations, automatically determining whether currently identified unsafe attitude behaviors occur within the compliance context constraints of railway operations. An early warning closed-loop management module generates tiered early warning events based on safety control rules for behaviors deemed unsafe, pushing them to the corresponding monitoring terminal. The module receives and records the feedback information, forming a closed-loop management mechanism throughout the entire process.

[0032] The aforementioned real-time monitoring system, method, device, computer equipment, computer-readable storage medium, and computer program products for railway workers' safety posture have achieved end-to-end technical integration from video stream access, lightweight edge inference, scenario-based model management, contextual fusion and discrimination to hierarchical early warning and closed-loop management. This is achieved by constructing a four-in-one real-time monitoring system for railway workers' safety posture, integrating video access edge analysis, intelligent recognition capabilities, multi-dimensional business fusion and linkage, and intelligent early warning closed-loop management. On the one hand, it completes low-latency, high-efficiency localized analysis of posture behavior at the edge, effectively utilizing existing video resources. On the other hand, relying on the railway operation compliance context quadruple and posture-rule mapping mechanism, it significantly improves the accuracy of identifying typical unsafe behaviors such as bending over, running, and abnormal standing posture, greatly reducing the false alarm rate. Simultaneously, through no-code training interaction and cloud-edge collaborative strategy distribution, on-site maintenance personnel can independently complete model iteration and strategy updates, breaking free from dependence on vendor-closed architectures and truly realizing adaptive evolution of intelligent analysis capabilities and full-process closed-loop management for complex working conditions of heavy-haul railways. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0034] Figure 1 This is a structural block diagram of a real-time monitoring system for the safe posture of railway workers in one embodiment;

[0035] Figure 2This is a structural block diagram of a real-time monitoring system for the safe posture of railway workers, as described in another embodiment.

[0036] Figure 3 This is a flowchart illustrating a method for real-time monitoring of the safe posture of railway workers in one embodiment;

[0037] Figure 4 This is a structural block diagram of a real-time monitoring device for the safe posture of railway workers in one embodiment;

[0038] Figure 5 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0040] In one embodiment, such as Figure 1 As shown, a real-time monitoring system 100 for the safe posture of railway workers is provided. The system includes: a video access and analysis module 101, a railway operation posture recognition module 102, a business integration and linkage module 103, and an early warning closed-loop management module 104. The modules are connected to each other.

[0041] The video access and analysis module 101 is used to access video surveillance resources along the railway line and use a lightweight attitude recognition inference unit deployed on the edge side to perform inference analysis on the workers through the video stream obtained from the video surveillance resources.

[0042] The railway operation posture recognition module 102 is used to centrally manage posture monitoring data for railway operation positions and to train posture recognition units in railway scenarios based on the posture monitoring data.

[0043] The business integration and linkage module 103 is used to associate and match the attitude recognition results generated by the attitude recognition unit with the multi-dimensional context data of railway operations, and automatically determine whether the currently identified unsafe attitude behaviors of railway operations are within the scope of railway operation compliance context constraints.

[0044] The early warning closed-loop management module 104 is used to generate graded early warning events for behaviors judged as unsafe postures in railway operations according to safety control rules, push them to the corresponding monitoring post terminals, receive and record the obtained handling feedback information, and form a closed-loop management mechanism for the entire process.

[0045] Specifically, refer to Figure 1The railway worker safety posture real-time monitoring system 100 responds to the received posture real-time monitoring command. The video access and analysis module 101 accesses video surveillance resources along the railway line and uses a lightweight posture recognition inference unit deployed at the edge to perform inference analysis on the workers through the video stream obtained from the video surveillance resources. The railway operation posture recognition module 102 centrally manages the posture monitoring data for railway operation positions and trains posture recognition units in the railway scenario based on the posture monitoring data. The business integration and linkage module 103 associates and matches the posture recognition results generated by the posture recognition unit with the multi-dimensional context data of railway operations to automatically determine whether the currently identified unsafe posture behavior of railway operations is within the scope of railway operation compliance context constraints. The early warning closed-loop management module 104 generates graded early warning events for behaviors determined to be unsafe postures of railway operations according to safety control rules, pushes them to the corresponding monitoring position terminals, receives and records the received handling feedback information, and forms a full-process closed-loop management mechanism.

[0046] The aforementioned Real-time Safety Posture Monitoring System 100 for Railway Workers integrates video access edge analysis, intelligent recognition capabilities, multi-dimensional business fusion and linkage, and intelligent early warning closed-loop management. This system achieves seamless integration across the entire technology chain, from video stream access, lightweight edge inference, scenario-based model management, contextual fusion discrimination, to tiered early warning and closed-loop handling. On one hand, it performs low-latency, high-efficiency localized posture and behavior analysis at the edge, effectively utilizing existing video resources. On the other hand, relying on the railway operation compliance context quadruple and posture-rule mapping mechanism, it significantly improves the accuracy of identifying typical unsafe behaviors such as bending over, running, and abnormal standing postures, greatly reducing the false alarm rate. Simultaneously, through no-code training interaction and cloud-edge collaborative strategy distribution, on-site maintenance personnel can independently complete model iteration and strategy updates, eliminating dependence on vendor-closed architectures and truly achieving adaptive evolution of intelligent analysis capabilities and full-process closed-loop management for complex heavy-haul railway conditions.

[0047] In one embodiment, the video access analysis module 101 includes an access adaptation submodule, a containerized operation submodule, and a posture inference execution submodule. The access adaptation submodule is used to be compatible with video surveillance resources along the railway line and uniformly convert them into standardized video streams. The containerized operation submodule is used to build an edge containerized environment based on integrated security management on the edge side. The posture inference execution submodule is used to call the deployed lightweight posture recognition inference unit in the edge containerized environment based on the standardized video stream, to perform real-time analysis of the spatiotemporal characteristics of the human posture of the workers in the standardized video stream, and output the identification results of unsafe posture behaviors in railway operations.

[0048] Specifically, the video access and analysis module 101 achieves end-to-end processing from video access to intelligent analysis through three levels of collaborative sub-modules: First, the access adaptation sub-module interfaces with multi-vendor, multi-protocol camera equipment deployed along the railway line, performs protocol parsing, format normalization, and bitstream optimization on the raw video stream, and outputs a unified standard video stream to ensure seamless access to heterogeneous monitoring resources by the system; second, the containerized operation sub-module 102 builds a lightweight container operating environment on the edge-side integrated security management equipment, dynamically scheduling and monitoring the health status of computing, storage, and network resources. It provides an isolated, stable, and manageable execution platform for intelligent analysis tasks. Finally, the posture inference execution submodule 103 loads the deployed lightweight posture recognition inference unit in this container environment to perform frame-by-frame analysis on the workers in the standardized video stream, extract their spatiotemporal features of human posture in real time, and identify unsafe posture behaviors in railway operations such as bending over, running, and abnormal standing postures. It outputs structured recognition results, thereby completing low-latency and high-efficiency localized intelligent analysis on the edge side, avoiding the need to send the original video back to the center, and effectively improving the system response speed and bandwidth utilization efficiency.

[0049] For example, the video access and analysis module 101, through three closely cooperating sub-modules, realizes a complete processing chain from heterogeneous video access to edge intelligent analysis: First, the access adaptation sub-module is responsible for connecting to various video surveillance devices already deployed along the railway line, ensuring compatibility with different manufacturers and different protocols (such as GB / T). The system utilizes camera outputs from 28181, ONVIF, RTSP, etc., to uniformly decode, adapt frame rates, and standardize formats of the raw video streams, generating standardized video streams with consistent structure and stable timing, providing high-quality input for subsequent analysis. Secondly, the containerized operation submodule constructs a lightweight, isolated container operating environment on the edge-side integrated security management equipment, supporting on-demand allocation and dynamic scheduling of CPU, memory, and AI acceleration resources. Simultaneously, it monitors container status, resource usage, and service health in real time, ensuring high availability and maintainability of inference tasks. Based on this, the posture inference execution submodule loads a lightweight posture recognition inference unit issued from the central side and adapted to railway operation scenarios. Within the container environment, it processes the workers in the standardized video stream frame by frame, extracting their key human body points, joint angles, and motion trajectories in real time. Based on preset rules, it identifies unsafe posture behaviors in railway operations such as bending over, running, and abnormal standing postures, outputting structured recognition results with timestamps, location information, and behavior types. This enables low-latency, low-bandwidth-dependent localized intelligent analysis at the edge, effectively supporting subsequent business linkage and early warning response.

[0050] In one embodiment, the railway operation posture recognition module 102 includes a behavior standard definition submodule, an intelligent training interaction submodule, and a version management submodule. The behavior standard definition submodule is used to define the categories and discrimination rules of unsafe posture behaviors for posture monitoring of railway operation positions, providing a standardized supervision basis for the training of posture recognition intelligent analysis units. The intelligent training interaction submodule is used to provide a user interface, supporting on-site operation and maintenance personnel to perform intelligent training task arrangement on target posture behaviors based on collected railway operation video clips without programming capabilities, and calling the underlying training engine to generate or optimize posture recognition intelligent analysis units suitable for line conditions. The version management submodule is used to perform intelligent analysis and full-cycle governance of multiple posture recognition intelligent analysis units trained in the railway scenario, and through linkage with the containerized operation submodule, completes the push and update of specified version analysis units to the edge side for integrated security management.

[0051] Specifically, the railway operation posture recognition module 102 defines the categories and judgment rules of unsafe posture behaviors for posture monitoring of railway operation positions through the behavior standard definition submodule, providing a standardized supervision basis for the training of posture recognition intelligent analysis units; it provides a user interface through the intelligent training interaction submodule, supporting on-site operation and maintenance personnel to perform intelligent training task arrangement on target posture behaviors based on collected railway operation video clips without programming capabilities, and calling the underlying training engine to generate or optimize posture recognition intelligent analysis units suitable for line conditions; and the version management submodule performs intelligent analysis and full-cycle governance of multiple posture recognition intelligent analysis units trained in the railway scenario, and completes the push and update of specified version analysis units to the edge side for integrated safety management through linkage with the edge containerized operation submodule.

[0052] In one embodiment, the business integration and linkage module 103 includes: a context data access submodule, an association rule engine submodule, and a compliance judgment submodule; the context data access submodule is used to connect to the multi-source business and spatial information system of railway operations, collect and process the four-tuple of railway operation compliance context, and form a unified format of railway operation multi-dimensional context data stream; the association rule engine submodule is used to build a logical rule library, map and bind the types of unsafe posture behaviors of railway operations with the corresponding compliance context constraints, and dynamically adjust the rule parameters through the management interface; the compliance judgment submodule is used to synchronously obtain the current railway operation multi-dimensional context data after receiving the recognition result output by the posture reasoning execution submodule, perform matching logic judgment based on the mapping relationship in the posture context association rule engine, output a binary judgment result of whether the target behavior is within the scope of compliance context constraints, and transmit the judgment result together with the railway operation multi-dimensional context data to the early warning closed-loop management module.

[0053] Specifically, the business integration and linkage module 103 achieves deep integration and intelligent judgment of posture recognition results with railway operation business context through three organically connected sub-modules: First, the context data access sub-module connects in real time with multiple source business and spatial information systems such as the construction operation plan system, personnel positioning system, line mileage database, and GIS geographic information system, extracting information such as track window operation time, electronic fence of operation area, job responsibility configuration, and real-time personnel location information, and integrating them into a unified format of "railway operation compliance context quadruple", forming a continuously updated multi-dimensional context data stream; Second, the association rule engine sub-module constructs a configurable logical rule library based on railway safety operation procedures, associating various unsafe posture behaviors in railway operations (such as bending over, running, and abnormal standing posture of safety personnel) with corresponding compliance context constraints. The system precisely maps and binds rules, for example, stipulating that "bending over is considered compliant only when performed during designated track opening hours, within designated track areas, and by maintenance personnel." A visual management interface allows maintenance personnel to adjust rule parameters as needed, ensuring consistency between the rules and actual site conditions. Finally, after receiving the posture recognition results from the lightweight posture reasoning execution submodule, the compliance judgment submodule simultaneously acquires the current multi-dimensional context data stream. Based on the pre-set mapping relationships in the rule engine, it performs joint matching and logical judgment on the four dimensions of time, space, position, and identity of the behavior, outputting a binary judgment result (compliant or non-compliant) indicating whether the behavior falls within the compliance context constraints. This judgment, along with the original recognition data, is then transmitted to the intelligent early warning closed-loop management module, providing accurate and reliable decision-making basis for subsequent tiered early warning and handling.

[0054] Furthermore, the compliance determination submodule uses the following compliance determination formula for binary determination:

[0055]

[0056] In the above formula, Indicates the compliance assessment result. Indicates the first one extracted from the current context Dimensional actual value, The first element in the railway operation compliance context quadruple represents the first element. A set of constraints.

[0057] In one embodiment, the early warning closed-loop management module 104 includes: a strategy configuration submodule, an alarm push submodule, and a handling tracking submodule; the strategy configuration submodule is used to define three-level linkage response strategy tuples corresponding to different combinations of unsafe posture behaviors and compliance context judgment results of railway operations based on railway safety control rules, and dynamically adjust the early warning handling strategy template through a visual interface to form an executable hierarchical early warning strategy library; the alarm push submodule is used to, after receiving the violation judgment result output by the compliance judgment submodule, match the corresponding early warning responsibility mapping pair according to the hierarchical early warning strategy library, and synchronously push the current early warning event to the monitoring post terminal through the message middleware; the handling tracking submodule is used to record the full-link operation log of each early warning event, periodically summarize unclosed events and trigger the overdue supervision mechanism.

[0058] Specifically, the early warning closed-loop management module 104 constructs a complete early warning and feedback closed loop from risk assessment to handling and archiving through three collaborative sub-modules: First, the strategy configuration sub-module, based on railway safety control rules, defines a three-level linkage response strategy tuple containing early warning level, response time limit, and responsible monitoring position for different types of unsafe posture behaviors in railway operations (such as running, bending over, abnormal standing posture) and their corresponding compliance judgment results (such as "violation" or "suspected violation"). It also supports managers in flexibly adjusting the early warning trigger conditions and handling process templates for various behaviors through a visual configuration interface, forming a structured and executable hierarchical early warning strategy library. Second, after receiving the violation judgment result output by the real-time compliance judgment sub-module, the alarm push sub-module immediately queries the strategy library to match the corresponding early warning strategy. The alarm-responsibility mapping system synchronously pushes structured early warning events, including video clips, behavior types, location coordinates, timestamps, and contextual judgment criteria, to various terminal channels of designated monitoring positions via message middleware. These channels include mobile apps, web monitoring consoles, voice broadcasting systems, and on-site audible and visual alarm devices, ensuring that alarm information reaches the responsible personnel in a timely and accurate manner. Finally, the handling and tracking submodule records the entire operation log of each early warning event from generation, push, receipt, manual confirmation to final handling feedback. It supports monitoring personnel to update the status of each event on the terminal, such as "handled," "false alarm marked," or "escalated report," and automatically summarizes and reminds those events that have not been handled within the time limit. This achieves a closed-loop management mechanism that is traceable, auditable, and assessable throughout the entire process from risk identification, early warning response to handling and archiving.

[0059] In one embodiment, such as Figure 2As shown, the real-time monitoring system for safe posture of railway workers 100 also includes: a cloud-edge collaborative strategy distribution module 105; the cloud-edge collaborative strategy distribution module 105 is used to package the latest posture recognition intelligent analysis unit and early warning and handling strategy template generated by the version management submodule and distribute them to the corresponding edge-side integrated safety control, so as to complete the update of the posture recognition intelligent analysis unit and the three-level linkage response strategy tuple, and synchronously transmit them back to the railway operation posture recognition module.

[0060] Specifically, the real-time monitoring system for safe posture of railway workers 100 constructs a cloud-edge collaborative strategy distribution module 105 between the video access analysis module 101 and the railway operation posture recognition module 102 using collaborative technical feature units. The cloud-edge collaborative strategy distribution module 105 is used to package and distribute the latest posture recognition intelligent analysis unit and the corresponding early warning and handling strategy template generated by the version management submodule to the corresponding edge-side integrated safety control, so as to complete the update of the posture recognition intelligent analysis unit and the three-level linkage response strategy tuple, and synchronously transmit it back to the railway operation posture recognition module, forming a self-iterative full-process closed-loop management mechanism.

[0061] Preferably, to further streamline the collaborative link between central-side intelligent model management and edge-side real-time analysis execution, this application introduces a cloud-edge collaborative strategy distribution submodule 105 between the video access analysis module 101 and the railway operation posture recognition module 102. This submodule is responsible for receiving the latest posture recognition intelligent analysis unit, which has been trained and verified in actual railway scenarios, output by the version management submodule, and the early warning-response strategy template generated by the strategy configuration submodule. It packages these two components into a unified edge deployment package and distributes it to the corresponding edge-side integrated safety management equipment based on the line section, operation type, and edge node identifier. After receiving the deployment package, the edge side seamlessly updates its local lightweight posture recognition inference unit and three-level linkage response strategy tuple through a hot-loading mechanism, without interrupting the video analysis service, ensuring that the edge intelligence capability remains synchronized with the current operation specifications and on-site conditions. Meanwhile, the entire chain of data generated by the edge during operation, including identification results, compliance judgment conclusions, and handling feedback, is transmitted back to the railway operation posture intelligent recognition capability module through a closed-loop data channel. This is used to automatically evaluate key indicators such as model accuracy and false alarm rate. When the performance is not up to standard, the system can trigger a new round of sample collection and no-code training tasks, thereby forming an adaptive closed-loop mechanism of "central training optimization, edge deployment execution, data feedback verification, and dynamic strategy iteration," realizing the continuous evolution and autonomous optimization of railway operation safety posture monitoring capabilities.

[0062] In one embodiment, such as Figure 3As shown, a method for real-time monitoring of the safe posture of railway workers is provided. Taking the application of this method to the real-time monitoring system 100 for the safe posture of railway workers as described in the above embodiment as an example, the method includes the following steps:

[0063] Step S301: Access video surveillance resources along the railway line through the video access and analysis module, and use the lightweight attitude recognition inference unit deployed on the edge side to perform inference analysis on the workers through the video stream obtained from the video surveillance resources.

[0064] Step S302: The posture monitoring data for railway operation positions is centrally managed through the railway operation posture recognition module. Based on the posture monitoring data, posture recognition units are trained in the railway scenario.

[0065] Step S303: The attitude recognition result generated by the attitude recognition unit is associated and matched with the multi-dimensional context data of railway operation by the business integration and linkage module, and the unsafe attitude behavior of railway operation currently identified is automatically determined to be within the scope of railway operation compliance context constraints.

[0066] Step S304: The early warning closed-loop management module generates graded early warning events for behaviors deemed as unsafe postures in railway operations based on safety control rules, pushes them to the corresponding monitoring post terminals, receives and records the obtained handling feedback information, and forms a closed-loop management mechanism for the entire process.

[0067] Specifically, the Real-time Safety Posture Monitoring System 100 for Railway Workers first accesses video surveillance resources along the railway line through a video access and analysis module. Utilizing a lightweight posture recognition inference unit deployed at the edge, it performs inference analysis on workers based on the video streams obtained from the video surveillance resources. Then, the railway work posture recognition module centrally manages posture monitoring data for railway work positions. Based on this data, a posture recognition unit is trained within the railway scenario. Next, the business fusion and linkage module correlates and matches the posture recognition results generated by the posture recognition unit with multi-dimensional contextual data of railway operations, automatically determining whether the currently identified unsafe posture behavior occurs within the compliance context constraints of railway operations. Finally, the early warning closed-loop management module generates tiered early warning events based on safety control rules for behaviors deemed unsafe postures in railway operations, pushes them to the corresponding monitoring terminal, receives and records the received feedback information, forming a full-process closed-loop management mechanism.

[0068] For specific limitations regarding the above steps, please refer to the relevant embodiments of the real-time monitoring system for the safe posture of railway workers, which will not be repeated here.

[0069] The aforementioned real-time monitoring method for the safe posture of railway workers integrates video access edge analysis, intelligent recognition capabilities, multi-dimensional business fusion and linkage, and intelligent early warning closed-loop management into a unified real-time monitoring system. This system achieves seamless integration across the entire technology chain, from video stream access, lightweight edge inference, scenario-based model management, contextual fusion discrimination, to tiered early warning and closed-loop handling. On one hand, it performs low-latency, high-efficiency localized analysis of posture behavior at the edge, effectively utilizing existing video resources. On the other hand, relying on the railway operation compliance context quadruple and posture-rule mapping mechanism, it significantly improves the accuracy of identifying typical unsafe behaviors such as bending over, running, and abnormal standing posture, greatly reducing the false alarm rate. Simultaneously, through no-code training interaction and cloud-edge collaborative strategy distribution, on-site maintenance personnel can independently complete model iteration and strategy updates, breaking free from dependence on vendor-closed architectures and truly achieving adaptive evolution of intelligent analysis capabilities and full-process closed-loop management for complex working conditions of heavy-haul railways.

[0070] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0071] Based on the same inventive concept, this application also provides a real-time monitoring device for the safe posture of railway workers, used to implement the aforementioned method for real-time monitoring of the safe posture of railway workers. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the real-time monitoring device for the safe posture of railway workers provided below can be found in the limitations of the real-time monitoring method for the safe posture of railway workers described above, and will not be repeated here.

[0072] In one exemplary embodiment, such as Figure 4As shown, a real-time monitoring device for the safe posture of railway workers is provided. Taking the application of this device in the real-time monitoring system 100 for the safe posture of railway workers as described in the above embodiment as an example, the device may include: an inference analysis unit 401, a scene training unit 402, an association matching unit 403, and a hierarchical early warning unit 404. The inference analysis unit 401 may be deployed in the video access analysis module as described in the above embodiment, the scene training unit 402 may be deployed in the railway operation posture recognition module as described in the above embodiment, the association matching unit 403 may be deployed in the business fusion linkage module as described in the above embodiment, and the hierarchical early warning unit 404 may be deployed in the early warning closed-loop management module as described in the above embodiment.

[0073] The reasoning and analysis unit 401 is used to access video surveillance resources along the railway line through the video access and analysis module, and to use the lightweight attitude recognition reasoning unit deployed on the edge side to perform reasoning and analysis on the operators through the video stream obtained from the video surveillance resources.

[0074] The scenario training unit 402 is used to centrally manage the attitude monitoring data for railway operation positions through the railway operation attitude recognition module, and to form an attitude recognition unit in the railway scenario training based on the attitude monitoring data.

[0075] The association matching unit 403 is used to use the business integration and linkage module to associate and match the attitude recognition results generated by the attitude recognition unit with the multi-dimensional context data of railway operations, and automatically determine whether the currently identified unsafe attitude behavior of railway operations is within the scope of railway operation compliance context constraints.

[0076] The graded early warning unit 404 is used to generate graded early warning events based on safety control rules for behaviors deemed as unsafe postures in railway operations using the early warning closed-loop management module, push them to the corresponding monitoring terminal, receive and record the obtained handling feedback information, and form a closed-loop management mechanism for the entire process.

[0077] Specific limitations regarding the real-time monitoring device for the safe posture of railway workers can be found in the limitations of the real-time monitoring method for the safe posture of railway workers described above, and will not be repeated here. Each module in the aforementioned real-time monitoring device for the safe posture of railway workers can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0078] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 5As shown, the computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When executed by the processor, the computer program implements a method for real-time monitoring of the safe posture of railway workers. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0079] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0080] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0081] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0082] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0083] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0084] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0085] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0086] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A real-time monitoring system for the safe posture of railway workers, characterized in that, The system includes: a video access and analysis module, a railway operation posture recognition module, a business integration and linkage module, and an early warning closed-loop management module; among which... The video access and analysis module is used to access video surveillance resources along the railway line and use a lightweight attitude recognition inference unit deployed on the edge side to perform inference analysis on the workers through the video stream obtained from the video surveillance resources. The railway operation posture recognition module is used to centrally manage posture monitoring data for railway operation positions, and to train posture recognition units in railway scenarios based on the posture monitoring data. The business integration and linkage module is used to associate and match the posture recognition results generated by the posture recognition unit with the multi-dimensional context data of railway operations, and automatically determine whether the currently identified unsafe posture behavior of railway operations is within the scope of railway operation compliance context constraints. The early warning closed-loop management module is used to generate graded early warning events for behaviors judged as unsafe postures in railway operations according to safety control rules, push them to the corresponding monitoring post terminals, and receive and record the obtained handling feedback information.

2. The system as described in claim 1, characterized in that: The video access analysis module includes an access adaptation submodule, a containerized operation submodule, and a posture inference execution submodule. The access adaptation submodule is used to be compatible with video surveillance resources along the railway line and uniformly convert them into standardized video streams; The containerized operation submodule is used to build an edge containerized environment based on integrated security management and control at the edge. The posture inference execution submodule is used to call the deployed lightweight posture recognition inference unit in the edge containerized environment based on the standardized video stream, to perform real-time analysis of the spatiotemporal features of the human posture of the workers in the standardized video stream, and output the identification results of unsafe posture behaviors in railway operations.

3. The system according to claim 2, characterized in that, The railway operation posture recognition module includes a behavior standard definition submodule, an intelligent training interaction submodule, and a version management submodule; The behavior standard definition submodule is used to define the categories and discrimination rules of unsafe posture behaviors for posture monitoring of railway operation positions, and to provide a standardized supervision basis for the training of the posture recognition intelligent analysis unit. The intelligent training interaction submodule is used to provide a user interaction interface, enabling on-site operation and maintenance personnel to perform intelligent training task arrangement on target posture behavior based on collected railway operation video clips without programming capabilities, and call the underlying training engine to generate or optimize the posture recognition intelligent analysis unit suitable for line working conditions. The version management submodule is used to perform intelligent analysis and full-cycle governance of multiple attitude recognition intelligent analysis units trained in the railway scenario. By linking with the containerized operation submodule, it can push and update the specified version analysis unit to the edge-side integrated security management.

4. The system according to claim 3, characterized in that, The business integration and linkage module includes: a context data access submodule, an association rule engine submodule, and a compliance judgment submodule; The context data access submodule is used to connect with the railway operation multi-source business and spatial information system, collect and process railway operation compliance context quadruples, and form a unified format railway operation multi-dimensional context data stream. The association rule engine submodule is used to build a logical rule library, map and bind unsafe posture behavior types of railway operations with corresponding compliance context constraints, and dynamically adjust rule parameters through the management interface; The compliance judgment submodule is used to synchronously acquire the railway operation multidimensional context data at the current time after receiving the recognition result output by the attitude reasoning execution submodule, perform matching logic judgment based on the mapping relationship in the attitude context association rule engine, output a binary judgment result on whether the target behavior is within the scope of compliance context constraints, and transmit the judgment result together with the railway operation multidimensional context data to the early warning closed-loop management module.

5. The system according to claim 4, characterized in that, The early warning closed-loop management module includes: a strategy configuration submodule, an alarm push submodule, and a handling tracking submodule; The strategy configuration submodule is used to define a three-level linkage response strategy tuple corresponding to different combinations of unsafe posture behaviors and compliance context judgment results in railway operations based on railway safety management rules. The early warning and handling strategy templates are dynamically adjusted through a visual interface to form an executable hierarchical early warning strategy library. The alarm push submodule is used to push the current alarm event to the monitoring terminal through the message middleware after receiving the violation judgment result output by the compliance judgment submodule and matching the corresponding alarm responsibility mapping pair according to the hierarchical alarm strategy library. The handling and tracking submodule is used to record the full-link operation log of each early warning event, periodically summarize unclosed events and trigger the overdue supervision mechanism.

6. The system according to claim 5, characterized in that, The system also includes: a cloud-edge collaborative strategy distribution module; The cloud-edge collaborative strategy distribution module is used to package the latest attitude recognition intelligent analysis unit and early warning and handling strategy template generated by the version management submodule and distribute them to the corresponding edge-side integrated security management and control, so as to complete the update of the attitude recognition intelligent analysis unit and the three-level linkage response strategy tuple, and synchronously transmit them back to the railway operation attitude recognition module.

7. A method for real-time monitoring of the safe posture of railway workers, characterized in that, The method, applied to the real-time monitoring system for the safe posture of railway workers as described in any one of claims 1 to 6, comprises: The video access and analysis module accesses video surveillance resources along the railway line, and uses a lightweight attitude recognition inference unit deployed at the edge to perform inference analysis on the video stream obtained from the video surveillance resources for the workers. The posture monitoring data for railway operation positions is centrally managed through the railway operation posture recognition module, and posture recognition units are trained and formed in railway scenarios based on the posture monitoring data. The business integration and linkage module is used to associate and match the posture recognition results generated by the posture recognition unit with the multi-dimensional context data of railway operations, and automatically determine whether the currently identified unsafe posture behavior of railway operations is within the scope of railway operation compliance context constraints. The early warning closed-loop management module generates graded early warning events based on safety control rules for behaviors deemed as unsafe postures in railway operations. These events are then pushed to the corresponding monitoring terminal, where the received feedback information is received and recorded, forming a closed-loop management mechanism for the entire process.

8. A real-time monitoring device for the safe posture of railway workers, characterized in that, The device, applied to the real-time monitoring system for the safe posture of railway workers as described in any one of claims 1 to 6, comprises: The reasoning and analysis unit is used to access video surveillance resources along the railway line through the video access and analysis module, and to perform reasoning and analysis on the workers by using the video stream obtained from the video surveillance resources through the lightweight attitude recognition reasoning unit deployed on the edge side. The scenario training unit is used to centrally manage the posture monitoring data for railway operation positions through the railway operation posture recognition module, and to form a posture recognition unit in the railway scenario based on the posture monitoring data. The association matching unit is used to use the business integration and linkage module to associate and match the posture recognition results generated by the posture recognition unit with the multi-dimensional context data of railway operations, and automatically determine whether the currently identified unsafe posture behavior of railway operations is within the scope of railway operation compliance context constraints. The graded early warning unit is used by the early warning closed-loop management module to generate graded early warning events for behaviors that are judged to be unsafe postures in railway operations according to safety control rules, and push them to the corresponding monitoring terminal. The unit receives and records the feedback information of the handling, forming a closed-loop management mechanism for the entire process.

9. 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 method of claim 7.

10. 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 method of claim 7.

Images