Rail transit machine room intelligent operation and maintenance method and system based on digital twinning

Abstract

The application discloses a kind of based on digital twinning rail traffic communication machine room intelligent operation and maintenance method and system, belong to rail transit system digital twinning and three-dimensional visualization technical field, system in intelligent operation and maintenance server includes data acquisition module, three-dimensional modeling module, digital twinning engine module and database module, intelligent operation and maintenance client includes dynamic loading scheduling module, real-time data binding module and visual interactive module;The system is rendered to the three-dimensional model of communication machine room by lightweight Web3D engine on-demand dynamic loading, in combination with digital twinning technology realizes the real-time perception of the running state of communication equipment, data synchronization, fault alarm and remote operation and maintenance.The application can realize multi-level model hierarchical loading, alarm driven state rendering, operation and maintenance closed-loop management and other functions.Compared with prior art, the application has the advantages of light model, fast response, easy deployment and strong scalability, and is widely applicable to communication infrastructure operation and maintenance scenarios in the field of rail transit.

Description

A method and system for intelligent operation and maintenance of rail transit communication equipment rooms based on digital twins Technical Field

[0001] This invention relates to an intelligent operation and maintenance method and system for rail transit communication equipment rooms based on digital twins, belonging to the field of digital twin and 3D visualization technology for rail transit systems, specifically involving an intelligent visualization operation and maintenance method and system for communication equipment rooms that integrates digital twins and lightweight Web3D. Background Technology

[0002] The reliability requirements for communication systems in urban rail transit are extremely high. With the expansion of the network, the number and types of equipment in communication equipment rooms have surged, and the coupling between subsystems has deepened, posing unprecedented challenges to the real-time, refined, and intelligent operation and maintenance management.

[0003] Currently, the operation and maintenance models in this field have mainly gone through the following stages, but all of them have significant limitations:

[0004] 1. Traditional Two-Dimensional Operation and Maintenance Model: Currently, most operation and maintenance systems are still based on a two-dimensional plane, presented in the form of device lists, topology diagrams, static rack diagrams, etc. For example, Chinese patent application CN109474479A, "A Network Device Monitoring Method and System," demonstrates a typical two-dimensional monitoring scheme, which collects data through the SNMP protocol and displays device status in a two-dimensional interface. The fundamental drawback of this method is that the information is not intuitive. Operation and maintenance personnel find it difficult to quickly correlate abstract alarm data with specific devices, boards, and even ports in the physical space of the data center, resulting in low efficiency in fault location. Especially in emergency situations, valuable handling time is wasted on information conversion and searching.

[0005] 2. Basic 3D Visualization Mode: To overcome the shortcomings of 2D systems, some advanced systems have begun to introduce 3D modeling technology in an attempt to construct virtual scenes of the computer room. However, these implementation schemes generally suffer from key technological shortcomings;

[0006] Model redundancy and loading performance bottlenecks: Directly using high-precision industrial models without optimization results in bloated model files, leading to slow loading in web browsers and severely limiting the system's usability and promotional value. This problem is particularly prominent in large-scale, multi-datacenter scenarios like rail transit.

[0007] Data and models are isolated: 3D models are often only used as static scene displays, lacking deep binding and driving mechanisms with real-time generated operational data (status, performance, alarms). The model and data are in an independent state, without establishing a dynamic correlation and mapping relationship, making it impossible to build a data-driven digital twin, which greatly reduces the application value of 3D scenes.

[0008] Weak interactive capabilities: Interactions are mostly limited to basic scene rotation and zooming, failing to achieve seamless drilling from the macro data center to the micro port, and also lacking the ability to automatically locate alarms and visualize interactive control based on real-time data.

[0009] In summary, existing technologies either remain at the abstract, non-spatial level or, while introducing 3D visualization, are hampered by performance, data, and interaction limitations, failing to form a comprehensive "digital twin" operation and maintenance system capable of handling real-time data, supporting fine-grained interaction, and suitable for lightweight Web deployment. Therefore, developing a system and method that integrates lightweight Web3D technology with the digital twin concept to achieve holographic mapping, real-time status synchronization, and intelligent operation and maintenance interaction of communication equipment rooms from the overall structure down to the port-level devices has become a pressing technical challenge in this field. Summary of the Invention

[0010] In view of the problems of model redundancy, loading performance bottlenecks, data and model isolation, and weak interaction capabilities in existing technologies, this invention constructs an intelligent operation and maintenance method and system for rail transit communication equipment rooms based on digital twins. Its core technical solution is: by constructing a lightweight, hierarchical 3D model of the communication equipment room, and relying on a digital twin engine to achieve dynamic mapping and interaction between the physical system and the virtual model, combined with real-time 3D display driven by equipment operating status, a rapid-response, status-linked, and data-driven visual operation and maintenance platform is constructed. The system uses a lightweight Web3D engine to dynamically load and render the 3D model on demand. By integrating multiple modules such as data acquisition, 3D modeling, dynamic loading scheduling, digital twin engine, database, real-time data binding, and visual interaction, it achieves real-time perception of the operating status of communication equipment, data synchronization, fault alarms, and remote operation and maintenance. The system has functions such as multi-level model hierarchical loading, alarm-driven status rendering, and closed-loop operation and maintenance management. Compared with existing technologies, it has advantages such as lightweight models, rapid response, ease of deployment, and strong scalability, and is widely applicable to the operation and maintenance scenarios of communication infrastructure in the rail transit field.

[0011] The technical solution adopted in this invention is: an intelligent operation and maintenance method for rail transit communication equipment rooms based on digital twins, implemented based on an intelligent operation and maintenance server and an intelligent operation and maintenance client. The intelligent operation and maintenance server includes a data acquisition module, a 3D modeling module, a digital twin engine module, and a database module. The intelligent operation and maintenance client includes a dynamic loading and scheduling module, a real-time data binding module, and a visualization interaction module. The specific steps are as follows:

[0012] Step 1: The intelligent operation and maintenance client's visualization interaction module starts when the user accesses the system URL through a browser. The visualization interaction module requests basic data such as site information, data center information, and external communication subsystem equipment information from the intelligent operation and maintenance server's database module. The basic data is then handed over to the intelligent operation and maintenance client's dynamic loading and scheduling module. Based on the user's initial perspective, the dynamic loading and scheduling module requests the required model resources from the intelligent operation and maintenance server's 3D modeling module. After obtaining the corresponding model resources, the dynamic loading and scheduling module hands them over to the visualization interaction module for 3D rendering.

[0013] Step 2: The data acquisition module uses an IoT gateway to continuously collect multi-source heterogeneous data from external communication subsystem devices. After parsing the collected data through multiple protocol interfaces such as COBRA, SNMP, Modbus, and HTTP, it sends the data to the digital twin engine module. The digital twin engine module maps the received data from the external communication subsystem devices to the corresponding 3D model in the virtual world according to preset rules, updates the current state of the digital twin, and stores historical data in the database module.

[0014] Step 3: The digital twin engine module sends real-time data to the real-time data binding module of the intelligent operation and maintenance client via WebSocket communication. After receiving the real-time data and parsing the data packet, the real-time data binding module locates the corresponding model object in the rendered 3D scene according to the device, subrack, slot, and port identifiers in the data, and drives the status update, information mounting, and alarm location of the 3D model.

[0015] Step 4: The visualization interaction module captures the user's interaction with the 3D scene, generates instructions, and sends the instructions to the digital twin engine module via API requests. This module verifies and processes the instructions. For query requests, it directly returns relevant information. For control commands, it sends control instructions to the communication subsystem device through the data acquisition module. After the communication subsystem device executes the operation, its state changes are recaptured by the data acquisition module and re-enter the digital twin engine module to send the updated real-time data of the digital twin to the real-time data binding module of the intelligent operation and maintenance client for 3D rendering and display on the interface.

[0016] Step 5: The 3D modeling module creates, optimizes, and hierarchically organizes the 3D model, and stores the generated lightweight model file on the intelligent operation and maintenance server for download by the dynamic loading and scheduling module of the intelligent operation and maintenance client.

[0017] The 3D modeling module uses GLB / GLTF 2.0 format for its 3D model resources, combined with Draco compression format and supporting LOD switching. It is divided into a six-level structure: room, cabinet, equipment, subrack, board, and port. Based on the distance between the model and the camera, it switches between models of different precision to optimize loading performance.

[0018] The core of the digital twin engine module lies in a hierarchical, scalable, unified twin data model. This model adopts an object-oriented design approach, defining a device base class containing common attributes. This base class includes a unique identifier, physical asset identifier, name, type, model ID, status, parent device ID, location information, and business attributes. Based on this base class, specific twin classes for data centers, cabinets, devices, subracks, slots, and ports are derived. Each twin class extends its unique set of attributes; the data center twin extends environmental monitoring attributes; the cabinet twin extends power and space attributes; the device twin extends computing resource attributes; the slot twin extends board status attributes; and the port twin extends network attributes.

[0019] The unified twin data model maintains a complete containment hierarchy tree from the data center to the port through the parentId and children list.

[0020] The dynamic loading and scheduling module continuously monitors three triggering conditions:

[0021] (1) User operations, including clicking and dragging;

[0022] (2) Changes in the field of view, including camera movement, rotation, and zoom;

[0023] (3) Status changes occur, including alarm triggering and device status updates;

[0024] When any condition is triggered, the system first determines whether it is a user operation; if so, it enters the corresponding decision branch to identify the resource target to be loaded; if not, the system will further determine whether it is triggered by a change in the field of view; if so, it enters the corresponding decision branch to identify the resource target to be loaded; if not, the system will further determine whether it is triggered by a change in state; if so, it enters the corresponding decision branch to identify the resource target to be loaded; if not, the current process ends.

[0025] When any of the above conditions is triggered, the system will enter a decision branch to determine the specific trigger type and identify the resource target that needs to be loaded accordingly:

[0026] (1) If the area division is triggered: the system identifies and determines the specific physical or logical area to which the user's current operation is directed, i.e. a specific computer room or cabinet. If the area division is not triggered, the hierarchical structure is judged.

[0027] (2) If the hierarchical structure is triggered: the system analyzes the tree-like hierarchical structure of the scene and determines the target level, that is, the user initiates the focusing operation from the "computer room" level → "rack" level → "equipment" level → "subrack" level → "slot" level → "port" level, from the upper level to the lower level in sequence until the target level is located. If the hierarchical structure is not triggered, the view frustum judgment is performed.

[0028] (3) If the view frustum judgment is triggered: the system calculates based on the view frustum of the camera and performs collision detection with the bounding box of the 3D model in the scene to accurately determine the current visible area. If the view frustum judgment is not triggered, the interactive behavior judgment is performed.

[0029] (4) If triggered by an interactive behavior: the system determines the specific interactive behavior and its associated object, namely the clicked device model and its detailed information panel. If the interactive behavior is not triggered, the process ends.

[0030] After the target resource is identified, the system will perform two operations simultaneously:

[0031] (1) Loading resources: Based on the judgment result, request and load model resources from the determined area, target level, view frustum or related to the interactive behavior from the intelligent operation and maintenance server;

[0032] (2) Unload resources: At the same time, the system will identify and actively unload model resources that are no longer needed:

[0033] i. Models in other regions not currently in use;

[0034] ii. Redundant models at non-target levels;

[0035] iii. Models located outside the current view frustum;

[0036] iv. Models unrelated to the current interaction behavior;

[0037] After completing the loading and unloading of resources in this round, the system instructs the Web3D rendering engine to execute scene rendering and present the latest model to the user.

[0038] The visualization interaction module includes 3D rendering and human-computer interaction components. The 3D rendering implements an on-demand dynamic loading rendering method. 3D rendering is performed when the rendering rules are met; otherwise, the rendering loop stops. On-demand rendering rules include successful model loading, mouse click, double click, right-click events, window changes, element size changes, switching perspectives, clearing the scene, and using smooth animation effects. The human-computer interaction component enables users to perform operations such as left-click dragging, moving, zooming, double-clicking to open a door, and right-clicking to close a door.

[0039] A digital twin-based intelligent operation and maintenance system for rail transit communication equipment rooms is disclosed. The system consists of two parts: an intelligent operation and maintenance client and an intelligent operation and maintenance server. Several intelligent operation and maintenance clients are connected to the intelligent operation and maintenance server via network cables and switches. The two communicate with each other through WebSocket and HTTP.

[0040] The beneficial technical effects of this invention are as follows: 1. It achieves high-performance web-based 3D visualization, fundamentally overcoming loading and rendering bottlenecks. By adopting compression formats such as glb / Draco and combining them with multi-level LOD (Level of Detail) technology, the model size is reduced by an average of over 70%. The dynamic loading scheduling module loads resources on demand based on the viewpoint and interaction behavior, ensuring that the first-screen loading time for large-scale data center scenes is controlled within 3 seconds, and maintaining a smooth interactive frame rate (≥30fps) on mainstream browsers, thus solving the core pain points of model size redundancy and low loading efficiency in basic 3D modes.

[0041] 2. Deep, real-time fusion of data and 3D models has been achieved, realizing a leap from "visible" to "knowable and manageable." By establishing a WebSocket real-time data channel and a precise identifier mapping mechanism, this system precisely binds device status (e.g., online / offline), performance indicators (e.g., CPU utilization, port traffic), and alarm information to specific devices, subracks, boards, and even ports in the virtual scene at the pixel level. Status update latency is less than 1 second, faulty devices are highlighted with a flashing red light, and the system automatically switches perspectives for location, completely changing the problem of data and spatial location being disconnected in traditional 2D or static 3D systems, improving fault location efficiency by over 80%.

[0042] 3. A panoramic, multi-dimensional monitoring view from logical topology to physical space has been constructed. This invention not only displays the physical layout of the data center, but also overlays the logical topology of communication links onto a three-dimensional space, achieving unified monitoring in both logical and physical dimensions. Maintenance personnel can intuitively see which specific cabinets, devices, and ports a service link passes through. When a link alarm occurs, all relevant physical devices can be quickly located in three-dimensional space, enabling cross-system correlation analysis and greatly improving the ability to troubleshoot complex problems.

[0043] 4. The system possesses high flexibility and scalability. Its modular design (such as separate client and server components and an independent digital twin engine) and region- and hierarchy-based dynamic loading rules enable smooth expansion to adapt to new lines, data centers, or equipment types. This loosely coupled architecture reduces the cost of later maintenance and upgrades, and its scalability is several times higher than that of traditional tightly coupled monolithic systems, demonstrating promising engineering application prospects. Attached Figure Description

[0044] Figure 1 is a system architecture diagram for implementing the present invention;

[0045] Figure 2 is a flowchart of the dynamic loading process of the present invention;

[0046] Figure 3 is a diagram of the binding and twin synchronization of device status data in the external communication subsystem of the present invention. Detailed Implementation

[0047] The present invention will be further described below with reference to the accompanying drawings and embodiments:

[0048] Figure 1 shows the system architecture diagram for implementing this invention: a smart operation and maintenance system for rail transit communication equipment rooms based on digital twins. The system consists of two parts: a smart operation and maintenance client and a smart operation and maintenance server. Several smart operation and maintenance clients are connected to the smart operation and maintenance server via network cables and switches, and the two communicate with each other through WebSocket and HTTP. Each smart operation and maintenance client includes a real-time data binding module, a visualization interaction module, and a dynamic loading and scheduling module; the smart operation and maintenance server includes a digital twin engine module, a 3D modeling module, a data acquisition module, and a database module.

[0049] Example 1: The intelligent operation and maintenance client is deployed in the control center of the rail transit subway line and the station work room, while the intelligent operation and maintenance server is deployed in the communication room of the control center. External communication subsystem equipment includes a network security system, transmission system, dedicated wireless system, passenger information system, video surveillance system, clock system, broadcasting system, dedicated power supply system, dedicated telephone system, and official telephone system.

[0050] The specific implementation steps are as follows:

[0051] I. Visual Interaction Module: Upon the user accessing the system URL via a browser, the visual interaction module is activated. This module requests basic data such as site information, data center information, and external communication subsystem equipment information from the intelligent operation and maintenance server's database module. This basic data is then passed to the dynamic loading and scheduling module. Based on the user's initial perspective, the dynamic loading and scheduling module requests the necessary model resources from the intelligent operation and maintenance server. After obtaining the corresponding model resources, the dynamic loading and scheduling module hands them over to the visual interaction module for 3D rendering.

[0052] II. The data acquisition module uses an IoT gateway to continuously collect multi-source heterogeneous data from the communication subsystem devices. After parsing the collected data through various protocol interfaces such as COBRA, SNMP, Modbus, and HTTP, it sends the data to the digital twin engine module. The digital twin engine module maps the received physical device data to the corresponding 3D model in the virtual world according to preset rules, updating the current state of the digital twin. Simultaneously, historical data is stored in the database module. The core of the digital twin engine module lies in a hierarchical, scalable, unified twin data model. This model adopts an object-oriented design approach, defining a device base class containing common attributes. All specific device types (computer room, cabinet, device, subrack, slot, port) inherit from this base class and possess extended attributes. See Table 1, Unified Metadata Model (Device Base Class), for the device base class.

[0053] Table 1 Unified Metadata Model (Device Base Class)

[0054]

[0055] Based on this base class, specific twin classes are derived for data center, rack, equipment, subrack, slot, and port. See Table 2 for data center twin classes, Table 3 for rack twin classes, Table 4 for equipment twin classes, Table 5 for subrack twin classes, Table 6 for slot twin classes, and Table 7 for port twin classes. Each class extends its unique set of attributes. For example, the data center twin extends environmental monitoring attributes (temperature and humidity); the rack twin extends power and space attributes (number of CPUs, power consumption); the equipment twin extends computing resource attributes (CPU, memory utilization); the slot twin extends board status attributes; and the port twin extends network attributes (status, traffic, optical power). The unified twin data model maintains a complete containment hierarchy tree from data center to port through a parentId and children list, ensuring accurate digital mapping between physical space and logical relationships. This design not only comprehensively describes the static attributes of the equipment, but also provides a structured data foundation for real-time data-driven visualization, topology analysis, and intelligent operation and maintenance decision-making through dynamic attributes. For example, the scenario description of the transmission system equipment in the external communication subsystem is as follows:

[0056] Data center: Line1_Core_Room;

[0057] Server rack: Rack A-12, 42U high;

[0058] Equipment: One optical transmission network backbone device is installed in position 15-20U of the cabinet;

[0059] Sub-rack: The third sub-rack of this equipment;

[0060] Slot: Slot number 5 of this sub-rack;

[0061] Port: The first optical port on this slot board, Port-1 (10G);

[0062] See Tables 2-7 for instance data of twins at each level.

[0063] Table 2: Computer Room Twin

[0064]

[0065] Table 3: Rack Twin

[0066]

[0067] Table 4: Equipment Twin

[0068]

[0069] Table 5: Subscaffold twins

[0070]

[0071] Table 6: Slot Twins

[0072]

[0073] Table 7: Port Twins

[0074]

[0075] 3. The digital twin engine module sends real-time data to the real-time data binding module of the intelligent operation and maintenance client via WebSocket communication. After receiving the data, the real-time data binding module parses the data packet and locates the corresponding model object in the rendered 3D scene according to the device, subrack, slot and port identifiers in it, and drives its status update, information attachment and alarm location.

[0076] (1) Status update: Normal devices are displayed in green, and faulty devices are displayed in red;

[0077] (2) Information mounting: Display a real-time performance parameter panel next to the equipment model;

[0078] (3) Alarm location: Automatically switch the viewpoint to the device that triggered the alarm to achieve rapid location.

[0079] Fourth, the visualization interaction module captures the user's interaction with the 3D scene, generates instructions, and sends the instructions to the digital twin engine module of the intelligent operation and maintenance server via API requests. This module verifies and processes the instructions. For query requests, it directly returns relevant information; for control commands, it sends control instructions to the communication subsystem devices through the data acquisition module. After the subsystem devices execute the operation, their state changes are recaptured by the data acquisition module and sent back to the digital twin engine module to send the data to the real-time data binding module of the intelligent operation and maintenance client for 3D rendering and display on the interface.

[0080] The 5.3D modeling module creates, optimizes, and hierarchically organizes 3D models, storing the generated lightweight model files on the intelligent operation and maintenance server for download by the dynamic loading and scheduling module of the intelligent operation and maintenance client. The 3D model resource format adopts GLB / GLTF 2.0, combined with Draco compression and supports LOD switching. It is divided into a six-level structure: room, rack, equipment, subrack, board, and port. Different model resolutions are switched based on the distance between the model and the camera to optimize loading performance.

[0081] Figure 2 shows the dynamic loading flowchart.

[0082] a. Process initiation and trigger determination;

[0083] The dynamic loading and scheduling module continuously monitors three main triggering conditions:

[0084] (1) User operations, including clicking and dragging;

[0085] (2) Changes in the field of view, including camera movement, rotation, and zoom;

[0086] (3) Status changes occur, including alarm triggering and device status updates;

[0087] When any condition is triggered, the system first determines whether it is a user operation; if so, it enters the corresponding decision branch to identify the resource target to be loaded; if not, the system will further determine whether it is triggered by a change in the field of view; if so, it enters the corresponding decision branch to identify the resource target to be loaded; if not, the system will further determine whether it is triggered by a change in state; if so, it enters the corresponding decision branch to identify the resource target to be loaded; if not, the current process ends.

[0088] b. Rule-based judgment and target region identification;

[0089] When any of the above conditions is triggered, the system will enter a decision branch to determine the specific trigger type and identify the resource target that needs to be loaded accordingly:

[0090] (1) If the area division is triggered: the system identifies and determines the specific physical or logical area (a specific computer room or cabinet) to which the user's current operation is directed; if the area division is not triggered, the hierarchical structure is judged.

[0091] (2) If the hierarchical structure is triggered: the system analyzes the tree-like hierarchical structure of the scene and determines the target level, that is, the user drills down level by level from the “computer room” level ---> “rack” level ---> “equipment” level ---> “subrack” level ---> “slot” level ---> “port” level; if the hierarchical structure is not triggered, then the view frustum judgment is performed;

[0092] (3) If the view frustum judgment is triggered: the system calculates based on the view frustum of the camera and performs collision detection with the bounding box of the 3D model in the scene to accurately determine the current visible area; if the view frustum judgment is not triggered, the interactive behavior judgment is performed.

[0093] (4) If the interaction is triggered: the system determines the specific interaction (clicking the device) and its associated object, namely the clicked device model and its detailed information panel; if the interaction is not triggered, the process ends.

[0094] c. Parallel processing of resource loading and unloading;

[0095] After the target resource is identified, the system will perform two operations simultaneously:

[0096] (1) Loading resources: Based on the judgment result, request and load model resources from the determined area, target level, view frustum or related to the interactive behavior from the intelligent operation and maintenance server;

[0097] (2) Unload resources: At the same time, the system will identify and actively unload model resources that are no longer needed:

[0098] i. Models in other regions not currently in use;

[0099] ii. Redundant models at non-target levels;

[0100] iii. Models located outside the current view frustum;

[0101] iv. Models unrelated to the current interaction behavior;

[0102] (3) Final rendering:

[0103] After completing the loading and unloading of resources in this round, the system instructs the Web3D rendering engine to execute scene rendering, presenting the latest and most necessary models to the user, and the process ends.

[0104] Figure 3 shows the binding and twin synchronization diagram of device status data in the external communication subsystem.

[0105] a. After the process starts, the system first determines and obtains the status data of the external communication subsystem devices. The system follows a decision-making logic:

[0106] Method 1: Receive active reports from external communication subsystem devices. If the system receives real-time status data reports initiated by physical devices, this data stream will be used; otherwise, active acquisition will be performed at set time intervals.

[0107] Method 2: Actively retrieve data at set time intervals. If active reporting is not triggered, the system will act as a client and initiate queries to the external communication subsystem device according to a preset time strategy (e.g., every 5 seconds) to actively retrieve the latest status data. If this condition is not met, the process will end.

[0108] b. After successfully acquiring the device status data from the external communication subsystem, the system performs a data-model binding operation. Regardless of the acquisition method, the data is sent to the digital twin engine. The engine uses the unique device identifier (device, subrack, slot, port information) carried in the data packet to find the corresponding digital twin model in the virtual scene.

[0109] c. After binding, the system performs state change judgment. The engine compares the newly received external communication subsystem device status data with the status currently recorded in the twin model to determine whether the operating status, performance indicators, or alarm information of the external communication subsystem device has changed.

[0110] d. Based on the changes and the resulting judgment, execute different synchronization strategies:

[0111] (1) If the data changes: the system will drive and update the corresponding visual attributes or data panel in the digital twin model according to the changed content. The model color of the faulty external communication subsystem device will be changed to red and flashing, and its latest CPU utilization will be updated in the information panel. This achieves precise and real-time synchronization of changes in the physical world to the virtual space;

[0112] (2) If the data does not change: the system will keep the current display state of the digital twin model unchanged and will not need to perform any update operations, thereby saving computing resources.

[0113] e. The process ends and waits to enter the next data acquisition and synchronization cycle.

Claims

1. A method for intelligent operation and maintenance of rail transit communication equipment rooms based on digital twins, characterized in that: This system is implemented using an intelligent operation and maintenance (O&M) server and an intelligent O&M client. The intelligent O&M server includes a data acquisition module, a 3D modeling module, a digital twin engine module, and a database module. The intelligent O&M client includes a dynamic loading and scheduling module, a real-time data binding module, and a visualization interaction module. The specific steps are as follows: Step 1: The intelligent O&M client's visualization interaction module, based on the user's access to the system URL via a browser, initiates the visualization interaction module and requests basic data such as site information, data center information, and external communication subsystem equipment information from the intelligent O&M server's database module. This basic data is then passed to the intelligent O&M client's dynamic loading and scheduling module for dynamic loading. The loading and scheduling module requests the required model resources from the 3D modeling module of the intelligent operation and maintenance server based on the user's initial perspective. After the dynamic loading and scheduling module obtains the corresponding model resources, it hands them over to the visualization interaction module for 3D rendering. In step two, the data acquisition module uses an IoT gateway to continuously collect multi-source heterogeneous data from external communication subsystem devices. After parsing the collected data through multiple protocol interfaces such as COBRA, SNMP, Modbus, and HTTP, it sends the data to the digital twin engine module. The digital twin engine module maps the received external communication subsystem device data one-to-one with the corresponding 3D model in the virtual world according to preset rules and updates the data accordingly. The current state of the digital twin is recorded, and historical data is stored in the database module. Step three: The digital twin engine module sends real-time data to the real-time data binding module of the intelligent operation and maintenance client via WebSocket communication. After receiving the real-time data and parsing the data packets, the real-time data binding module locates the corresponding model object in the rendered 3D scene based on the device, subrack, slot, and port identifiers, driving the 3D model's state update, information mounting, and alarm location. Step four: The visualization interaction module captures the user's interaction with the 3D scene, generates instructions, and sends the instructions to the digital twin engine module via API requests. The block verifies and processes the instructions. For query requests, it directly returns relevant information. For control commands, it sends control instructions to the communication subsystem device through the data acquisition module. After the communication subsystem device executes the operation, its state changes are recaptured by the data acquisition module and re-enter the digital twin engine module to send the updated real-time data of the digital twin to the real-time data binding module of the intelligent operation and maintenance client for 3D rendering and display on the interface. Step 5: The three-dimensional modeling module creates, optimizes, and hierarchically organizes the three-dimensional model and stores the generated lightweight model file on the intelligent operation and maintenance server for download by the dynamic loading and scheduling module of the intelligent operation and maintenance client.

2. The intelligent operation and maintenance method for rail transit communication equipment rooms based on digital twins according to claim 1, characterized in that: The 3D modeling module uses GLB / GLTF 2.0 format for its 3D model resources, combined with Draco compression format and supporting LOD switching. It is divided into a six-level structure: room, cabinet, equipment, subrack, board, and port. Based on the distance between the model and the camera, it switches between models of different precision to optimize loading performance.

3. The intelligent operation and maintenance method for rail transit communication equipment rooms based on digital twins according to claim 1, characterized in that: The core of the digital twin engine module lies in a hierarchical, scalable, unified twin data model. This model adopts an object-oriented design approach, defining a device base class containing common attributes. This base class includes a unique identifier, physical asset identifier, name, type, model ID, status, parent device ID, location information, and business attributes. Based on this base class, specific twin classes for data centers, cabinets, devices, subracks, slots, and ports are derived. Each twin class extends its unique set of attributes; the data center twin extends environmental monitoring attributes; the cabinet twin extends power and space attributes; the device twin extends computing resource attributes; the slot twin extends board status attributes; and the port twin extends network attributes. The unified twin data model maintains a complete containment hierarchy tree from data center to port through a parentId and children list.

4. The intelligent operation and maintenance method for rail transit communication equipment rooms based on digital twins according to claim 1, characterized in that: The dynamic loading scheduling module continuously monitors three triggering conditions: (1) user operation, including clicking and dragging; (2) change of viewing angle, including camera movement, rotation, and zoom; (3) state change, including alarm triggering and device state update. When any condition is triggered, the system first determines whether it is a user operation. If so, it enters the corresponding decision branch to identify the resource target to be loaded. If not, the system will further determine whether it is triggered by a change of viewing angle. If so, it enters the corresponding decision branch to identify the resource target to be loaded. If not, the system will further determine whether it is triggered by a state change. If so, it enters the corresponding decision branch to identify the resource target to be loaded. If not, the current process ends. When any of the above conditions are triggered, the system will enter a decision branch to determine the specific trigger type and identify the resource target to be loaded accordingly: (A) If it is a region division trigger: the system identifies and determines the specific computer room or cabinet to which the user's current operation is directed. If the region division is not triggered, the hierarchical structure is judged. (B) If it is a hierarchical structure trigger The system analyzes the tree-like hierarchical structure of the scene. The user initiates focusing operations sequentially from the "server room" level to the "rack" level, "equipment" level, "subrack" level, "slot" level, and "port" level, until the target level is located. If the hierarchical structure is not triggered, a view frustum judgment is performed. (C) If the view frustum judgment is triggered: the system calculates based on the camera's view frustum and performs collision detection with the bounding box of the 3D model in the scene to accurately determine the current visible area. If the view frustum judgment is not triggered, interaction is performed. Behavior judgment; (D) If triggered by interactive behavior: The system judges the clicked device model and its detailed information panel. If the interactive behavior is not triggered, the process ends; After determining the target resource, the system will perform two operations simultaneously: (a) Load resources: Based on the judgment result, request and load the model resources in the determined area, target level, view frustum or related to the interactive behavior from the intelligent operation and maintenance server; (b) Unload resources: At the same time, the system will identify and actively unload the model resources that are no longer needed: i. Other models that are not in the current area; ii. Redundant models at non-target levels; iii. Models outside the current view frustum; iv. Models unrelated to the current interaction behavior; After completing the loading and unloading of resources in this round, the system instructs the Web3D rendering engine to execute scene rendering and present the latest model to the user.

5. The intelligent operation and maintenance method for rail transit communication equipment rooms based on digital twins according to claim 1, characterized in that: The visualization interaction module includes 3D rendering and human-computer interaction components. The 3D rendering implements an on-demand dynamic loading rendering method. 3D rendering is performed when the rendering rules are met; otherwise, the rendering loop stops. On-demand rendering rules include successful model loading, mouse click, double click, right-click events, window changes, element size changes, switching perspectives, clearing the scene, and using smooth animation effects. The human-computer interaction component allows users to perform operations such as left-click dragging, moving, scaling, double-clicking to open a door, and right-clicking to close a door.

6. A smart operation and maintenance system for rail transit communication equipment rooms based on digital twins, characterized in that: The system consists of two parts: an intelligent operation and maintenance client and an intelligent operation and maintenance server. Several intelligent operation and maintenance clients are connected to the intelligent operation and maintenance server via network cables and switches, and the two communicate with each other through WebSocket and HTTP.

Images