Tunnel structure monitoring and risk visualization early warning system based on digital twinning

Abstract

The invention relates to the field of tunnel structure monitoring, in particular to a tunnel structure monitoring and risk visualization early warning system and method based on digital twinning. A data acquisition and transmission module; a BIM three-dimensional model module; a data fusion and analysis module; a risk assessment and early warning module; a visual display module; an event response and linkage module; according to the method, BIM digital twinning is taken as a unified carrier, multi-source sensing and components are bidirectionally bound, structural state prediction and anomaly recognition are realized through fusion and a depth model, and a partition grading threshold value and three-dimensional / AR visualization are established; after early warning is triggered, drainage, ventilation, illumination and other devices are linked to form a monitoring-early warning-disposal closed loop, so that the accuracy, timeliness and traceability of risk identification are improved.

Description

Technical Field

[0001] This invention relates to the field of tunnel structure monitoring, and specifically to a tunnel structure monitoring and risk visualization early warning system based on digital twins. Background Technology

[0002] As a crucial component of transportation infrastructure, the structural safety of tunnels directly impacts the smooth operation of public transportation and the safety of people and property. With rapid urbanization and the development of transportation networks, the number and scale of subway, highway, and railway tunnels are constantly increasing, highlighting the growing structural safety issues during tunnel construction and long-term operation. Therefore, conducting tunnel structural health monitoring and risk early warning has become an important research direction in infrastructure safety management.

[0003] Currently, tunnel structure monitoring mainly relies on two methods: manual inspection and sensor monitoring. Manual inspection typically assesses structural health by periodically checking for cracks, leaks, and deformations on the tunnel lining surface. However, this method suffers from problems such as high cyclicality, significant time lag, low efficiency, and subjective results, making it difficult to detect potential safety hazards in a timely manner. While sensor monitoring can continuously collect parameters such as tunnel stress, strain, displacement, temperature, humidity, and vibration, existing systems generally suffer from reliance on a single data source, data isolation, and insufficient data fusion, resulting in monitoring results that fail to accurately reflect the overall health status of the tunnel.

[0004] In existing technologies, some tunnel projects employ distributed fiber optic sensing technology or wireless sensor networks for structural health monitoring, achieving a certain degree of wide-area coverage and automated data collection. However, the results are mostly at the data level, lacking correlation with the tunnel's three-dimensional structural model and failing to intuitively reflect the tunnel's stress and deformation in the spatial dimension. Furthermore, existing risk assessment methods largely rely on static threshold settings or empirical judgments, making it difficult to handle dynamic evolution processes under complex working conditions. This leads to false alarms and missed alarms, limiting the reliability and applicability of early warning systems.

[0005] Furthermore, current tunnel monitoring systems generally lack a linkage mechanism with the operation and maintenance system. When a risk event occurs, the monitoring platform can often only issue alarm information, but cannot automatically link with the tunnel's drainage, ventilation, lighting, and other systems, resulting in a time lag between early warning and response, thereby reducing the efficiency of emergency response.

[0006] Chinese invention patent application CN112345678A discloses a tunnel lining structure monitoring system based on fiber optic sensing. This patent uses fiber optic grating sensors deployed in the tunnel lining to achieve long-term monitoring of strain and cracks; however, it relies solely on fiber optic data and lacks multi-source sensor fusion; it does not incorporate a BIM model, and the results are not intuitive.

[0007] Chinese invention patent application CN114987654A discloses a tunnel structure health monitoring and risk assessment system, which introduces a big data analysis model to cluster and detect anomalies in the monitoring data. However, its assessment model relies on historical data, lacks dynamic adaptability, and is not linked with the operation and maintenance system.

[0008] US Patent Application No. US20220123456A1 discloses a Tunnel monitoring and maintenance planning system, which provides maintenance decision support based on sensors and numerical models; however, the system focuses on maintenance planning and does not achieve real-time risk warning and dynamic display of zones.

[0009] Japanese invention patent application No. JP20230098765A discloses a tunnel operation monitoring and visualization system that combines tunnel operation data and sensor data to provide a two-dimensional visualization interface; however, it lacks three-dimensional BIM integration, can only display local monitoring information, and does not form a global risk visualization early warning system.

[0010] In summary, existing tunnel structure monitoring technologies have the following shortcomings: Information fragmentation: The lack of integration between sensor monitoring data and tunnel structure models makes it difficult to achieve a global and intuitive display. Algorithm shortcomings: Most existing risk assessments are based on empirical values ​​or single thresholds, which cannot dynamically adapt to complex and ever-changing working conditions. Delayed early warning: The early warning mechanism relies on manual intervention, lacks real-time capability, and is subject to the risk of false alarms and missed alarms. Lack of collaboration: The existing system is not well coupled with the tunnel's operation and maintenance subsystem and lacks emergency response and control capabilities.

[0011] Therefore, there is an urgent need for a new method for monitoring and visualizing risks of tunnel structures that can deeply integrate BIM 3D models with multi-source sensor monitoring data and combine intelligent algorithms to achieve dynamic graphical display of structural status, risk zoning early warning and event response linkage, so as to overcome the shortcomings of existing technologies. Summary of the Invention

[0012] Based on the above description, this invention provides a tunnel structure monitoring and risk visualization early warning system and method based on digital twins. Using BIM digital twins as a unified carrier, it bidirectionally binds multi-source sensors to components, and edge caching and breakpoint retransmission ensure data integrity and timeliness. Through fusion and deep modeling, it achieves structural state prediction and anomaly identification, establishes zoning and hierarchical thresholds and 3D / AR visualization; after an early warning is triggered, it links drainage, ventilation, lighting, and other equipment, forming a closed loop of monitoring-early warning-response, significantly improving the accuracy, timeliness, and traceability of risk identification throughout the entire lifecycle.

[0013] On the one hand, the technical solution of the present invention to solve the above-mentioned technical problems is as follows: a tunnel structure monitoring and risk visualization early warning system based on digital twin, comprising: a sensor monitoring module, which deploys various sensors in the tunnel and collects data with unified time stamp; and a data acquisition and transmission module, which buffers and retransmits multi-source data at the edge side and uploads it to the center through a secure link;

[0014] The BIM 3D model module builds and maintains a tunnel BIM 3D digital twin model, stores the geometric, material and maintenance attributes of components, and binds monitoring points to model components in two directions and supports observation-based dynamic updates.

[0015] The data fusion and analysis module performs denoising, interpolation, standardization, semantic alignment, and fusion on multi-source data. Based on a pre-trained deep learning model and combined with structural topology and working condition information, it identifies health status and predicts trends. The risk assessment and early warning module classifies and assesses risks in different zones based on the fusion results, sets multi-level trigger conditions, and generates disposal recommendations. The visualization module displays monitoring and risk information on the BIM 3D model using zone coloring, timeline playback, and component-level drill-down, and provides augmented reality on-site guidance to the terminal. The event response and linkage module implements parameterized linkage control of drainage, ventilation, lighting, and alarm devices and records closed-loop results when an early warning is triggered.

[0016] It should be understood that this system is applicable to both the construction and operation phases. It uses a digital twin model as a unified carrier for data fusion, visualization, and collaborative decision-making, thus avoiding the limitations of solutions that rely solely on a single sensing medium, a construction scenario, or a two-dimensional display.

[0017] Through a closed-loop architecture of "sensing—collection—twin—fusion—assessment—visualization—linkage," monitoring data is bidirectionally bound to BIM components, unifying data semantics and spatial positioning, and achieving consistent status perception and risk management across stages (construction and operation). Compared to two-dimensional display or single-media monitoring, this significantly improves the availability of multi-source data, the accuracy of risk identification, and the timeliness of response, forming a traceable full lifecycle management chain.

[0018] Based on the above technical solution, the present invention can be further improved as follows.

[0019] Furthermore, the BIM 3D model module uses an incremental update method to write monitoring-driven geometric and attribute changes into the digital twin, and records version evolution with a unique component identifier and a time window. The increment satisfies: ;in For a moment The twin state, Establish mapping relationships with components, measurement points, and processing versions for playback and traceability.

[0020] By employing the above technical solutions and using incremental updates and versioned records of twin models, changes in geometry and attributes are precisely located at the component and time window, achieving dynamic alignment that is "what you see is what you get." This improves the consistency between the model and the actual site, supports playback, auditing, and comparative analysis, and reduces misjudgments and delays in operation and maintenance decisions caused by model drift.

[0021] Furthermore, wherein the data acquisition and transmission module performs interpolation and standardization preprocessing and records data quality labels at the edge side, and the steps satisfy: , ;

[0022] It employs caching and breakpoint retransmission to maintain integrity, and uses encrypted channels and verification to maintain security and consistency.

[0023] By employing the aforementioned technical solutions, edge-side interpolation and standardized preprocessing, combined with caching, retransmission, and secure channels, ensure data continuity, integrity, and consistency, reducing the impact of communication jitter on analysis results. This strategy improves the quality and comparability of time-series data, providing stable input for subsequent fusion and intelligent evaluation, and reducing false positives and false negatives.

[0024] Furthermore, the data fusion and analysis module deeply integrates multi-source sensing and BIM component attributes to form health indicators and constructs a data lineage diagram of "component-measuring point-time window-version"; the fusion and indicator calculation satisfy: , ;

[0025] in For the first Source data, The value of is constrained by confidence level and spatiotemporal correlation.

[0026] The above technical solution generates health indicators through deep fusion and establishes a data lineage map of "component-measuring point-time window-version," enabling the measurement of the contribution and reliability of multi-source information within a unified feature space. The effect is improved robustness and interpretability of the indicators, facilitating rapid identification of anomaly sources and responsible components, and supporting refined maintenance decisions.

[0027] Furthermore, the data fusion and analysis module suppresses drift and jump noise based on state-space recursion and observation correction, and performs consistency comparison with the deep learning output; the recursive relationship satisfies: ; and with ;

[0028] The results are updated to obtain a corrected sequence of health indicators for subsequent zoning assessments.

[0029] By employing the aforementioned technical solutions, state-space recursion and observation correction are used to suppress drift and abrupt noise, and consistency comparisons are performed with the results of the deep model. This allows for stable evaluation under conditions of noise, missing measurements, and sensor offset. Consequently, the robustness of time series prediction and anomaly identification is improved, false triggers are reduced, and the ability to identify gradual degradation and sudden events is enhanced.

[0030] Furthermore, the risk assessment and early warning module sets multi-level thresholds and triggering strategies, providing alarm, early warning, and emergency response levels for components, zones, and sections respectively. When triggered, the event response and linkage module automatically adjusts the sampling frequency, drainage pump power, and ventilation speed according to the response table, and writes the response results and time-series trajectory back to the digital twin timeline. The triggering can be characterized by the following formula: ;

[0031] in As a current risk measure, This corresponds to the threshold.

[0032] Through the above technical solution, multi-level thresholds and triggering strategies express risks hierarchically from components and zones to cross-sections, and link them with sampling frequency, drainage, and ventilation parameters. This linkage enables an automatic transition from "problem detection" to "problem handling," shortening the average response time and reducing the duration of high-risk events and secondary losses.

[0033] Secondly, the technical solution of the present invention to solve the above-mentioned technical problems is as follows: a method for tunnel structure monitoring and risk visualization early warning based on digital twins, comprising:

[0034] S1, construct and initialize the BIM 3D digital twin model, and complete the two-way binding of monitoring points and model components and metadata registration;

[0035] S2, convergent displacement sensors, strain gauges, accelerometers, tilt sensors, and piezometers are deployed at key locations such as the lining, arch waist, arch crown, invert, and joints, and data are collected using a unified time scale. After edge buffering and breakpoint retransmission, the data is uploaded.

[0036] S3 performs denoising, interpolation, standardization, and semantic alignment, performs multi-source fusion, and outputs health status recognition and trend prediction based on a pre-trained deep learning model combined with structural topology.

[0037] S4. Based on the fusion results, risk classification and multi-level triggering judgment are performed in different areas to form disposal suggestions and linkage parameters;

[0038] S5 performs zoning and coloring, timeline playback, and component-level drill-down display on the BIM 3D model and augmented reality interface. When the trigger conditions are met, it implements parametric linkage control of drainage, ventilation, lighting, and alarm devices, records the closed loop, and writes back the twin version.

[0039] Through the above technical solutions, the methodology links modeling, deployment, preprocessing, fusion evaluation, and 3D / AR display into standardized operations, forming a reusable and auditable process specification. The effect is to translate disparate technical elements into executable steps, improving engineering implementation efficiency and cross-team collaboration efficiency, and ensuring consistency and transferability across different projects.

[0040] Furthermore, the deep learning prediction of S3 adopts a joint approach of spatial feature extraction and temporal modeling. First, the sensor and component features are encoded, and then the temporal correlation model is performed to obtain the state estimate and uncertainty range of the future window. The training samples consist of historical monitoring data, structural inspection results and maintenance records, and new data is used online for small-batch incremental updates.

[0041] Through the above technical solution, spatial feature extraction combined with temporal modeling and support for online mini-batch incremental updates enables the model to continuously self-calibrate as the environment and operating conditions evolve. This mechanism maintains prediction accuracy under data distribution drift, shortens the retraining window, and improves adaptability to new operating conditions and large-scale scenarios.

[0042] Furthermore, the time-series risk assessment of S4 employs a joint mapping of historical and real-time data and outputs partitioned quantification results, which can be characterized by the following formula: ,as well as: ;in For the current window data, For historical sample sets, These are the model parameters, and the results are used to determine the warning level and response priority.

[0043] Through the above technical solution, historical-real-time joint mapping and zoned quantitative output decouple overall risk from local risk, facilitating resource allocation based on regional priorities. The result is the achievement of risk prioritization and response scheduling that is "precise down to the zone level and detailed down to the component level," improving the return on investment for inspections and reinforcement.

[0044] Furthermore, the triggering and linkage of S5 are executed synchronously at the component level and the zone level: when the triggering condition is determined, the target component and adjacent components are highlighted in the BIM 3D model and the handling steps are pushed, the monitoring sampling frequency is automatically increased, the drainage pump power and ventilation speed are increased, and lighting and audible and visual alarms are linked when necessary; after the handling is completed, a handling report is generated, written into the digital twin timeline according to the component identification and time window, and made available for playback and auditing.

[0045] Through the above technical solution, component-level and zone-level processes are synchronized and linked, with overlaid highlighting of visual information and handling guidelines. Sampling, drainage, and ventilation are automatically enhanced, and a report is generated and written back to the timeline afterward. The effect is to achieve closed-loop governance of "visibility-controllability-verifiability," significantly improving the execution of handling procedures, accountability, review, and continuous improvement capabilities.

[0046] Compared with the prior art, the technical solution of this application has the following beneficial technical effects:

[0047] 1. Twin-driven multi-source fusion and accurate identification: Using BIM digital twins as a unified carrier, multi-source data such as stress, displacement, vibration, and seepage pressure are bound to individual components. Interpolation, standardization, and buffering are completed at the edge, while deep fusion and time-series prediction are performed at the center. The data is then displayed in a 3D / AR interface by zone and component level. This enables high-precision, interpretable, and traceable structural health identification and trend prediction across the entire construction-operation lifecycle, significantly reducing missed and false alarms and improving assessment robustness.

[0048] 2. A tiered early warning and coordinated response closed-loop system is established, creating multi-level thresholds and triggering mechanisms for components / zones / sections. Upon an early warning, parameterized controls for sampling frequency, drainage, ventilation, and lighting are automatically activated, and the response process and results are written back to a twin timeline and version record, forming a closed-loop governance system of "monitoring—assessment—early warning—execution—feedback." This mechanism reduces response time, minimizes high-risk continuations and secondary losses, and strengthens audit review and continuous optimization capabilities. Attached Figure Description

[0049] Figure 1 This is a schematic diagram of the overall system architecture of a tunnel structure monitoring and risk visualization early warning system based on digital twins according to Embodiment 1 of the present invention;

[0050] Figure 2 This is a functional structure diagram of the BIM 3D model module in Embodiment 1 of the present invention;

[0051] Figure 3 This is a functional structure diagram of the sensor monitoring module in Embodiment 1 of the present invention;

[0052] Figure 4 This is a functional structure diagram of the data acquisition and transmission module in Embodiment 1 of the present invention;

[0053] Figure 5 This is a functional structure diagram of the data fusion and analysis module in Embodiment 1 of the present invention;

[0054] Figure 6 This is a functional structure diagram of the risk assessment and early warning module in Embodiment 1 of the present invention;

[0055] Figure 7This is a functional structure diagram of the event response and linkage module of Embodiment 1 of the present invention;

[0056] Figure 8 This is a schematic diagram of the data flow and interaction in Embodiment 1 of the present invention;

[0057] Figure 9 This is a schematic diagram of the visual interface of Embodiment 1 of the present invention;

[0058] Figure 10 This is a flowchart of the risk assessment and early warning process in Embodiment 2 of the present invention;

[0059] Figure 11 This is a flowchart illustrating the event response and control linkage of Embodiment 2 of the present invention;

[0060] Figure 12 This is a flowchart illustrating the implementation steps of the method in Embodiment 2 of the present invention. Detailed Implementation

[0061] To facilitate understanding of this application, a more complete description will be provided below with reference to the accompanying drawings, which illustrate embodiments of the present application. However, the present application can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that the disclosure of this application will be thorough and complete.

[0062] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application.

[0063] Example 1: This example discloses a tunnel structure monitoring and risk visualization early warning system based on digital twins. This system integrates BIM 3D models with multi-source sensor monitoring data to achieve dynamic graphical display of structural status, intelligent fusion of multi-source data, risk zoning early warning, and event response linkage, thereby improving the health monitoring and safety assurance capabilities of tunnel structures throughout their entire lifecycle.

[0064] I. BIM 3D Model Module: In this embodiment, the BIM 3D model module first establishes an initial 3D digital model based on the structural data from the tunnel design phase. This model includes not only geometric information such as tunnel cross-sections and ring sections, but also embedded construction procedures, material parameters, and maintenance history. During operation, the module dynamically updates itself by linking with monitoring data, ensuring consistency between the digital model and the actual structure.

[0065] In terms of implementation, initial BIM models are generated using modeling software such as Revit and Navisworks, and real-time input from sensors and construction logs is received via data interfaces. Regarding display, the module supports multi-level data overlay and, combined with augmented reality or virtual reality technologies, achieves immersive visualization. This allows maintenance personnel to intuitively obtain the structural health status and achieve global situational awareness.

[0066] II. Sensor Monitoring Module: This module deploys various types of sensor nodes at key locations in the tunnel (including the arch crown, sidewalls, joints, and foundation) to collect parameters such as stress, strain, displacement, temperature, humidity, and vibration in real time. Specifically, convergent displacement sensors, strain gauges, accelerometers, tilt sensors, piezometers, and temperature and humidity sensors are deployed at the tunnel lining, arch waist, arch crown, invert, joints, and areas of concentrated deformation, and the data is collected using a unified timescale. Among these, stress and strain sensors reflect the structural stress level, displacement sensors monitor deformation development, temperature and humidity sensors analyze environmental impacts, and vibration sensors identify construction disturbances or seismic effects. Through optimized deployment and redundant configuration, these sensors form a comprehensive monitoring network, providing a continuous and reliable data foundation for subsequent system analysis.

[0067] III. Data Acquisition and Transmission Module: Raw signals acquired by sensors are initially processed by edge nodes and then transmitted to the central processing platform via wireless transmission (such as LoRa, NB-IoT, Wi-Fi) or wired transmission (such as fiber optic, Ethernet). To ensure data continuity and integrity, nodes are equipped with a caching mechanism to temporarily store data when the network is unstable and automatically retransmit it after recovery. This module enables efficient aggregation of monitoring data, providing real-time input for risk assessment.

[0068] IV. Data Fusion and Analysis Module: Within the central processing platform, this module preprocesses and fuses multi-source heterogeneous data, employing deep learning and machine learning algorithms to establish a structural health assessment model. Specifically, it utilizes multi-source data fusion algorithms to eliminate single-sensor errors, improving overall judgment accuracy; it analyzes structural evolution trends through time-series prediction models (such as LSTM and CNN-LSTM combined models); and it combines BIM models to generate a three-dimensional dynamic graphical display. In this way, maintenance personnel can not only monitor the health status in real time but also achieve proactive prevention and control through trend prediction.

[0069] V. Risk Assessment and Early Warning Module: This module establishes a multi-level risk assessment model based on historical monitoring data and set safety thresholds. When structural indicators approach or exceed the thresholds, the system automatically determines the risk level and maps the risk zones to the BIM model. The risk status of each area of ​​the tunnel is identified through color coding and early warning prompts. For example, when an anomaly occurs in the entrance section or joint area, the module can trigger a yellow or red warning and simultaneously push the results to the operation and maintenance platform, achieving real-time zoned early warning.

[0070] VI. Event Response and Linkage Module: When the system detects anomalies or early warning information, this module automatically triggers a response mechanism. For example, when the risk escalates, it increases the sampling frequency, notifies management personnel in real time, and links with other tunnel systems (such as drainage, ventilation, and lighting systems) for adjustments to reduce the risk level. In emergency situations, the module can directly activate emergency plans, including personnel evacuation and traffic control measures, thereby achieving closed-loop management of monitoring, early warning, and response.

[0071] BIM 3D model module is preferred;

[0072] 1. The modeling and updating of the BIM 3D model is as follows: In the implementation of the BIM 3D model module, the 3D digital model obtained in the tunnel design phase is first used as the initial basic data of the system. This BIM model is usually modeled using BIM software such as Revit and Navisworks, providing the tunnel's geometry, structural design information, and preliminary construction plan.

[0073] 1) Model building in the design phase: During the tunnel design phase, a three-dimensional digital model of the tunnel is created based on geological surveys, design drawings, and preliminary construction plans. This model includes the tunnel's external outline, structural components, support systems, tunnel locations, and other key architectural details.

[0074] 2) Dynamic Updates During Construction: During the construction phase, the BIM model integrates with on-site sensors and equipment interfaces to receive real-time data related to the construction status. This data can include construction progress, material usage, and the construction status of structural nodes. Whenever new data arrives, the BIM model automatically updates, reflecting the differences between the actual construction progress and the design model. For example, sensor data can provide information on structural deformation and settlement, and the model can display potential deformations and problems that may occur during construction.

[0075] 3) Real-time updates during the operation phase: Once the tunnel enters the operation phase, the BIM model connects to various installed sensors (such as strain, displacement, temperature, humidity, and vibration sensors) via data interfaces to dynamically receive real-time monitoring data. This data, combined with design information and construction records within the model, allows the BIM model to not only display the tunnel's geometry but also reflect its real-time structural health. The data acquisition module continuously updates the BIM model with real-time sensor data (such as stress changes or temperature fluctuations at a specific location within the tunnel), ensuring the model remains up-to-date and reflects changes in the tunnel's health.

[0076] Through this real-time data update-based approach, the BIM model becomes a dynamic tunnel management tool with real-time feedback, providing accurate visual data for subsequent monitoring, analysis, and maintenance.

[0077] 2. Multi-level Information Integration: Based on the BIM 3D model, this module integrates multiple types of information related to tunnel health status, ensuring that various tunnel data can be integrated and displayed on a unified platform. The integrated data not only includes traditional structural information but also covers data from different sensors and environmental monitoring, greatly improving the comprehensiveness and accuracy of tunnel health monitoring.

[0078] 1) Construction Phase Data Integration: Data collected during the construction process, including information on construction materials, process control, construction status, and geological changes, is integrated with the BIM model to update relevant data in the model in real time. For example, if the construction materials for a section of the tunnel change, or if a construction node is not completed as planned, the system will automatically identify and update the model to ensure that the model data is always consistent with the actual construction situation on site.

[0079] 2) Sensor Data Integration: Various sensors deployed within the tunnel (such as strain sensors, displacement sensors, temperature and humidity sensors, etc.) continuously collect data on environmental and structural changes. These sensors reflect the real-time health status of the tunnel structure, such as structural deformation, stress concentration, and temperature and humidity changes. Through data interfaces, this sensor data will be integrated into the BIM model to form a real-time updated structural health map.

[0080] 3) Environmental Data Integration: Changes in the environment surrounding the tunnel, including temperature and humidity, soil pressure, seismic impacts, and groundwater levels, can affect the tunnel's structural stability. Integrating real-time data from external environmental monitoring systems and combining these environmental factors with the tunnel's structural information helps provide a more accurate health assessment. For example, displacement changes in a section of the tunnel may be caused by variations in surrounding soil pressure; combining environmental data can help analyze the specific reasons for these structural changes.

[0081] By effectively integrating multi-level and multi-dimensional information such as construction data, sensor data, and environmental data into the BIM model, the system can display comprehensive tunnel data on the same platform, providing operation and maintenance personnel with comprehensive information support for efficient monitoring and analysis.

[0082] 3. Combination of Virtual and Reality: To enhance the visualization of tunnel health status, this module combines augmented reality (AR) and virtual reality (VR) technologies to improve the intuitiveness, accuracy, and interactivity of tunnel management and monitoring.

[0083] 1) Augmented Reality (AR) Technology: Using AR technology, maintenance personnel can scan the tunnel structure on-site using mobile devices (such as tablets, smartphones, AR glasses, etc.) to view the real-time overlay effect of the virtual BIM 3D model on the actual structure. Maintenance personnel can see the tunnel's health status, deformation data, monitoring results, and potential risk areas on the device screen. This technology provides on-site staff with an intuitive way to quickly assess the tunnel's health and identify problems promptly.

[0084] 2) AR Function Example: Suppose that a section of the tunnel has excessive stress or displacement. Maintenance personnel can directly scan the tunnel structure in that area using an AR device. The virtual stress / displacement layer will be superimposed on the actual tunnel structure and displayed on the device screen. Maintenance personnel can clearly see the anomaly in that area and take timely action.

[0085] 3) Virtual Reality (VR) Technology: In scenarios such as remote monitoring or remote training, maintenance personnel can utilize VR technology for immersive experiences. In a VR environment, maintenance personnel can enter the tunnel's virtual space, view its health status and monitoring data in 360 degrees, and conduct simulation tests or emergency response drills. Through this technology, maintenance personnel can gain a comprehensive understanding of the tunnel's health status and make informed decisions without actually being on-site.

[0086] 4) VR Function Example: Maintenance personnel can enter a virtual tunnel through VR equipment to simulate the impact of different environmental conditions (such as earthquakes, floods, etc.) on the tunnel structure, conduct virtual structural inspections, and even perform virtual maintenance operations.

[0087] By combining virtual and real-world methods, the health status of tunnels can be displayed not only in two-dimensional charts or static images, but also dynamically displayed and analyzed through interactive experiences in a real environment, greatly improving the effectiveness and accuracy of structural monitoring.

[0088] The BIM 3D model module in this embodiment achieves the following technical effects: ① Real-time dynamic updates: By integrating construction, sensor, and environmental data, the BIM model can be updated in real time, ensuring it always reflects the latest state of the tunnel. ② Comprehensive monitoring and analysis: Combining construction, sensor, and environmental data provides comprehensive structural health information, avoiding the limitations of a single data source. ③ Enhanced visualization and interactivity: Through the combination of AR and VR technologies, maintenance personnel can intuitively view the tunnel's health status on-site or remotely, and promptly identify potential risks. ④ Improved decision-making efficiency and accuracy: Real-time, dynamic BIM model updates combined with precise sensor data enhance risk prediction and decision support capabilities, making tunnel management more efficient and accurate.

[0089] Preferably, the sensor monitoring module employs stress / strain sensors, displacement sensors, temperature and humidity sensors, and vibration sensors arranged in the tunnel for monitoring. The specific sensor selection and arrangement are as follows:

[0090] 1. Sensor Selection 1) Stress / strain sensors are a key component of tunnel structure monitoring, capable of measuring the load and deformation borne by the tunnel structure in real time during construction or operation. Stress sensors monitor the tunnel's load-bearing capacity in real time, ensuring it remains within the designed safety range. Strain sensors are mainly used to detect minute deformations inside the tunnel, promptly identifying deformations or cracks exceeding the designed safety range, thereby preventing major accidents.

[0091] Stress sensor: Resistive stress sensors are typically used, and their working principle is based on the characteristic that the resistance of a material changes with stress. The sensor is attached to the surface of the structure and uses the resistance change caused by stress to calculate the stress distribution of the structure. Strain sensor: By employing fiber Bragg grating technology or capacitive strain sensors, tunnel deformation can be accurately detected by measuring optical or capacitive changes caused by minute deformations on the structural surface. Fiber Bragg grating sensors transmit signals through optical fibers, enabling effective remote monitoring and improving anti-interference capabilities.

[0092] Specifically, stress and strain sensors are installed on important parts of the tunnel, such as the roof, sidewalls, and connecting nodes, to monitor the stress distribution and deformation of the tunnel under load in real time, especially during the tunnel construction and operation phases, to ensure the safety of the structure at different stages.

[0093] 2) Displacement sensor, Functions and applications: Displacement sensors are primarily used to monitor the deformation or displacement of tunnel structures, especially changes in horizontal, vertical, and tilt displacement. Through high-precision measurements, displacement sensors provide crucial data support for the timely detection of potential structural problems. Key monitoring areas include tunnel joints, supporting structures, and surrounding rock boundaries. Laser rangefinder:Displacement can be accurately detected by measuring the change in distance between the tunnel surface and the laser emission source using a laser beam. eddy currents Flow displacement sensor: By utilizing the principle of eddy currents, the displacement of tunnel components can be accurately measured by sensing minute displacement changes on the surface of a metal object. Fiber optic displacement sensor: By using optical fiber to transmit signals, and based on the principle of light refraction or reflection, precise displacement can be measured by detecting changes in light intensity.

[0094] Specifically, displacement sensors are installed in key parts of the tunnel, such as walls, joints, and supporting structures, to monitor in real time whether the structure has excessive displacement and to avoid structural instability caused by deformation.

[0095] 3) Temperature and humidity sensors are used to monitor temperature and humidity changes within the tunnel and assess the potential impact of environmental conditions on the tunnel structure. Especially in underground tunnels or humid environments, changes in temperature and humidity can cause expansion, contraction, or corrosion of structural materials. Therefore, monitoring temperature and humidity changes is crucial for ensuring the long-term safe operation of the tunnel.

[0096] Semiconductor temperature and humidity sensor: It uses the principle of semiconductor material changes to sense changes in ambient temperature and humidity. By changing the conductivity of the sensor material, temperature and humidity are measured in real time. Capacitive temperature and humidity sensor: It utilizes the principle of capacitance change, where changes in ambient humidity cause changes in capacitance, which in turn senses changes in humidity.

[0097] Specifically, these devices are installed in tunnel interior walls, ventilation systems, and moisture-leaking areas to monitor changes in temperature and humidity. Through data feedback, they help analyze the potential impact of environmental factors on the tunnel structure, such as stress concentration characteristics caused by temperature and humidity.

[0098] 4) Vibration sensors monitor tunnel structural vibrations caused by external factors (such as earthquakes, construction vibrations, traffic loads, etc.), and can detect key parameters such as vibration frequency, amplitude, and period. These sensors are of great significance for monitoring tunnel health, especially during tunnel construction, repair, or when subjected to external influences; timely detection of abnormal vibrations is crucial to avoiding catastrophic structural damage.

[0099] Accelerometer: The vibration of a structure is measured using the principle of acceleration variation, and the acceleration signal is converted into an electrical signal for transmission and analysis. Piezoelectric sensor: Piezoelectric sensors monitor vibrations by generating changes in electrical charge under pressure or vibration, converting pressure signals into processable electrical signals.

[0100] Specifically, vibration sensors are installed at both ends of the tunnel, in the construction area, and in areas with heavy traffic to detect the tunnel's response to vibration sources in real time, so as to respond to abnormal vibrations and avoid structural damage caused by excessive vibration.

[0101] 2. Sensor Deployment and Integration

[0102] 1) Sensor Deployment Locations: To ensure comprehensive monitoring coverage of all parts of the tunnel, sensors should be deployed at key structural locations, including but not limited to: ① Tunnel Walls: Monitor stress, displacement, and temperature / humidity changes in the walls, especially potential issues such as tunnel wall thickness and cracking. ② Support Structures: Focus on monitoring the stress, displacement, and deformation of load-bearing structures such as support columns and frames to ensure their long-term stability. ③ Joints: Tunnel connection nodes and joints are areas prone to cracking and stress concentration; sensors need to be densely deployed in these areas. ④ Floor and Roof: The floor and roof structures need to be monitored for settlement, displacement, and stress, especially when the tunnel floor is subjected to settlement or upward pressure, to ensure structural safety.

[0103] 2) Data Acquisition and Integration: Sensors transmit collected data to data centers or cloud platforms via wireless or wired communication technologies for real-time monitoring and analysis. During data transmission, efficient and stable communication protocols (such as Zigbee, LoRa, 5G, and Wi-Fi) are employed to ensure data real-time performance and stability.

[0104] Data from all sensors is integrated and processed through a unified platform and updated in real time according to system requirements. The modular design allows for flexible selection and expansion of sensor types, and new sensors can be added to the system at any time to meet different tunnel monitoring needs.

[0105] 3) System Integration and Expansion: The integration of sensors and the system utilizes open interface technology, ensuring seamless interoperability between sensors of different brands and types. The system is designed with an open API interface, facilitating integration with existing tunnel monitoring systems and supporting future technology or sensor upgrades and expansions.

[0106] The sensor monitoring module in this embodiment achieves the following technical effects: ① Comprehensive and accurate health monitoring: By deploying various types of sensors, the system can monitor environmental changes inside and outside the tunnel in real time, ensuring a comprehensive understanding of structural health and providing accurate data support for subsequent risk analysis, health diagnosis, and decision-making. ② Real-time data transmission and analysis: Real-time data transmission and analysis ensure the timeliness and accuracy of tunnel monitoring, enabling rapid response when structural or environmental changes occur, reducing potential safety hazards. ③ Enhanced risk warning capability: Real-time monitoring and feedback of sensor data can promptly capture potential risks, such as structural deformation and stress exceeding limits, providing accurate basis for risk assessment and early warning mechanisms. ④ Scalability and flexibility: The system supports flexible integration and expansion of sensors, allowing for the addition of new monitoring points or upgrading of existing sensors according to actual needs, ensuring the system has the capability for long-term operation and technological updates. ⑤ Improved operation and maintenance efficiency and safety: Sensors monitor the health status of the tunnel in real time, allowing maintenance personnel to perform timely maintenance based on real-time data, reducing manual inspection work and improving safety.

[0107] Preferably, the data acquisition and transmission module transmits data collected by various sensors to the central processing platform wirelessly or via wired means, ensuring the real-time nature, integrity, and reliability of the data during transmission. The core function of this module is to transmit real-time data collected by tunnel structure and environmental monitoring sensors to a data center or cloud platform, providing efficient and stable data support for subsequent analysis, evaluation, and decision-making. Through efficient communication protocols and a reliable data transmission system, the data acquisition and transmission module ensures the stable operation of the tunnel monitoring system in complex environments and can supplement and recover data in the event of network instability or data loss.

[0108] 1. Data transmission: Wireless transmission: This module employs advanced wireless transmission technology to transmit the collected sensor data to the data center in real time. Wireless transmission technologies include, but are not limited to, the following: Wi-Fi: Suitable for short-distance, high-bandwidth transmission scenarios, data transmitted via Wi-Fi protocol has high speed and low latency, which can meet the real-time requirements of most monitoring data. LoRa: Suitable for long-distance, low-power data transmission, LoRa networks are particularly well-suited for monitoring systems of large infrastructures such as tunnels. They support long-distance data transmission while maintaining low power consumption, making them ideal for environments where wired networks are difficult to deploy, such as tunnels. NB-IoT: Based on narrowband IoT technology, it is suitable for data transmission of large-scale IoT devices, can work stably in long-distance, low-bandwidth environments, and is suitable for long-term monitoring of the environment inside and outside tunnels. Wired transmission technology:When the sensor deployment area has good communication infrastructure, wired transmission is used to ensure efficient and stable data transmission. Commonly used wired transmission technologies include: Fiber optic transmission: Data collected by sensors is transmitted to a central data processing platform at high speed and stably via optical fiber. Optical fiber transmission features low latency and high bandwidth, making it suitable for large-scale data transmission and high-precision monitoring. Ethernet transmission: This technology utilizes traditional Ethernet communication protocols and employs wired connections to transmit data to a monitoring center or cloud platform. It is suitable for use in environments with existing network infrastructure, offering good stability and low network interference.

[0109] During data transmission, encryption technology is used to protect data privacy and security, ensuring that information is not tampered with or leaked during transmission.

[0110] 2. Data buffering and processing; Data buffer: To address network fluctuations or outages, this module incorporates a buffer in sensor nodes or relay devices. This buffer temporarily stores collected data, ensuring timely upload once the network recovers. The caching mechanism effectively prevents data loss due to network instability, guaranteeing data integrity. In Low Power Wide Area Network (LPWAN) environments such as LoRa or NB-IoT, if communication signals are unstable or the device is in a dead zone, the sensor can cache data locally and automatically upload it to the data center or cloud platform once the signal is restored. Data Department reason: Data transmitted to the central platform undergoes preliminary processing, including data cleaning, noise reduction, and format conversion. Through the data preprocessing module, the transmitted raw data is automatically standardized to eliminate noise interference and ensure data accuracy and reliability. The data processing module can also perform real-time data analysis and trend prediction to identify potential risks or abnormal changes.

[0111] 3. Communication Protocols and Interfaces: Communication between the sensors and the data acquisition platform is conducted through standardized protocols such as MQTT, HTTP, and CoAP. A unified data interface allows for seamless integration of data from sensors of different brands and types, ensuring the system's openness and scalability. MQTT protocol: Data is transmitted using the lightweight MQTT protocol, which offers significant advantages in low-bandwidth, low-power network environments. MQTT's publish / subscribe mechanism is suitable for real-time transmission of tunnel monitoring data, ensuring that sensor data reaches the monitoring platform in real time. HTTP / HTTPS protocol:In scenarios requiring secure and reliable data transmission, HTTP or HTTPS protocols are used for data exchange to ensure data security during transmission. The system features an open API, facilitating integration with other monitoring systems or databases and supporting future hardware or software expansion and upgrades.

[0112] 4. Redundancy and fault recovery Redundancy design: To ensure the system continues to operate normally in the event of a single point of failure, the data acquisition and transmission modules employ a redundant design. Data transmission lines and equipment (such as relay stations and data gateways) all support backup and switching to prevent data loss due to equipment failure. Fault detection and recovery mechanism: This module features automatic fault detection, enabling real-time monitoring of network communication status and anomalies during data transmission. Upon detecting a communication failure or data loss, the system automatically activates a fault recovery mechanism, ensuring data continuity and stability by reconnecting or switching to a backup channel.

[0113] 5. Real-time performance and low latency: Due to the high real-time requirements of the tunnel structure monitoring system, the data acquisition and transmission module utilizes efficient network communication technology, optimized data processing algorithms, and accelerated data transmission paths to ensure that data is promptly transmitted to the data center for real-time analysis after sensor acquisition. The system's latency is controlled at the millisecond level, meeting the needs for real-time monitoring and early warning of tunnel health status.

[0114] 6. System Integration and Scalability: In this embodiment, the data acquisition and transmission module uses an open interface (API) for seamless integration with the sensor, enabling it to adapt to different models and brands of sensors and supporting future technology expansion and upgrades. As the demand for tunnel monitoring continues to increase, the system can easily add new sensor types or data acquisition points, expanding the system's monitoring capabilities.

[0115] The data acquisition and transmission module in this embodiment has the following technical advantages: ① Ensuring real-time, integrity, and reliability of data transmission: It employs a combination of wireless and wired transmission to ensure efficient and stable data transmission in various environments. Simultaneously, caching mechanisms and redundancy design guarantee data integrity and transmission stability. ② Highly efficient data processing and analysis capabilities: Through efficient data transmission protocols and data processing modules, this module can process large amounts of sensor data promptly and accurately, providing real-time health status analysis and risk assessment. ③ Good system scalability and compatibility: The system supports integration with other monitoring systems through open interfaces and has excellent scalability, allowing for hardware and software upgrades as technology advances and needs change. ④ Ensuring system continuity and stability: The data acquisition and transmission module employs redundancy and fault recovery mechanisms to ensure stable system operation under various unforeseen circumstances, reducing potential interruptions or data loss during monitoring.

[0116] Preferably, the data fusion and analysis module is designed to intelligently fuse and analyze data from different sensors. Utilizing deep learning and machine learning algorithms, the module can assess and diagnose the health status of the tunnel structure and generate a dynamic graphical representation of the structure. This module, through efficient data fusion and analysis methods, reduces the errors and inaccuracies caused by single data sources, thereby improving the reliability and accuracy of the tunnel structure monitoring system.

[0117] The main functions of this module are: Intelligent Data Fusion: Deeply fusing various data collected from different types of sensors (such as stress, strain, displacement, temperature, humidity, vibration, etc.) to obtain a comprehensive health assessment. Structural Status Assessment and Diagnosis: Utilizing deep learning and machine learning algorithms to analyze the fused data, assessing the tunnel's health status, risk level, deformation trends, etc., providing a basis for subsequent decision-making. Dynamic Graphical Display: Combining the BIM 3D model, visualizing real-time monitoring data, intuitively displaying the tunnel's health status, and providing maintenance personnel with convenient monitoring tools.

[0118] The specific implementation method is as follows: 1. Data fusion algorithm. Data fusion is one of the core technologies of this module. By combining data from different sensors, the system can obtain a more comprehensive perspective and reduce the errors that may be caused by data from a single sensor. The fusion algorithm used is mainly based on deep learning and machine learning models, and the specific implementation method is as follows:

[0119] Multi-source data fusion based on deep learning: This invention employs deep learning algorithms (such as Convolutional Neural Networks (CNNs) and Long Short-Term Memory Networks (LSTMs)) to fuse data from different types of sensors. For example, displacement data, stress data, and environmental data (such as temperature and humidity) collected by sensors are preprocessed and then subjected to deep learning and multi-level feature extraction using a neural network model to obtain a highly accurate comprehensive evaluation result. This process effectively removes data noise and improves the accuracy of structural health status analysis.

[0120] The merged data: ;

[0121] in, This represents the merged data. It consists of data from different sensors. Deep learning algorithms use multi-layered neural networks to jointly analyze various types of data and identify potential structural risks.

[0122] Data denoising and standardization: Before data fusion, the raw data needs to be preprocessed, such as denoising and standardization. Data denoising methods include using techniques such as wavelet transform and Kalman filtering to eliminate interference signals that may be generated by sensors; standardization methods unify data from different sources to the same dimension, ensuring data comparability and consistency.

[0123] 2. Structural condition assessment: After data fusion, machine learning and deep learning algorithms are used to assess the health status of the tunnel. Commonly used machine learning models include Support Vector Machine (SVM), Decision Tree, and Random Forest. These models build predictive models of health status and structural risk by training on historical data.

[0124] Health status assessment: By training and learning from historical data, machine learning models can assess the health status of tunnels and provide risk levels (such as normal, warning, critical). For example, the model can determine whether the tunnel is within a safe range based on trends in stress, displacement, and vibration data.

[0125] Deformation trend prediction: By using the LSTM (Long Short-Term Memory) model in deep learning, the deformation trend of the tunnel in the future can be predicted to help determine whether potential structural risks will occur.

[0126] Risk grading: Based on the fused multidimensional data and combined with the evaluation algorithm, the tunnel is divided into risk zones. For example, an SVM model is trained to identify the risk level of each area of ​​the tunnel, and zone warnings are generated based on the predicted health status.

[0127] Risk level: ;in, Indicates the risk level. For the fused sensor data, For model parameters, This is the evaluation function. The risk level of the tunnel is generated through model calculations.

[0128] 3. Dynamic graphical display: To enhance the visualization of tunnel health monitoring, the data fusion and analysis module will integrate with the BIM 3D model to dynamically display real-time monitoring data. Through the visualization platform, maintenance personnel can intuitively view the health status of the tunnel structure and promptly identify potential risk areas. The specific implementation method is as follows:

[0129] Real-time BIM model updates: The BIM model is updated in real time based on sensor data, generating a dynamic graphical display of the tunnel's health status. The system uses integrated WebGL and Three.js graphics rendering technologies to render real-time data onto the 3D model, creating a visualization effect. For example, data such as tunnel wall deformation and stress concentration are displayed using color and transparency, clearly indicating healthy and hazardous areas.

[0130] Highlighting of Risk Areas: In the BIM 3D model, areas with different risk levels are highlighted using color or markers. For example, when the model calculates that a section of tunnel has a high risk, that area will be displayed in red to attract the attention of maintenance personnel and facilitate timely action.

[0131] User Interface: Maintenance personnel can select different monitoring indicators for real-time viewing through the interactive interface. For example, they can display data such as stress values ​​and displacement values ​​at a specific location, and view the tunnel's changing trends in real time through dynamic charts.

[0132] Augmented Reality (AR) / Virtual Reality (VR) Integration: Further enhances visualization effects. The system can combine the BIM model with the actual tunnel environment through AR / VR technology, allowing maintenance personnel to obtain an immersive health monitoring experience through mobile devices or VR headsets.

[0133] In this embodiment, the data fusion and analysis module has the following technical effects: ① Improved data analysis accuracy: By fusing and analyzing sensor data through deep learning and machine learning algorithms, data noise can be eliminated, improving the accuracy of health assessment. Compared with traditional single-source monitoring methods, this invention can comprehensively identify potential risks and make accurate predictions. ② Real-time risk assessment: The fused data provides an accurate and real-time basis for tunnel health assessment. Through risk assessment and zoned early warning, structural problems can be detected in a timely manner, avoiding serious accidents. ③ Enhanced visualization: Combining BIM 3D models and real-time monitoring data, this invention realizes a dynamic graphical display of tunnel health status. Maintenance personnel can use intuitive 3D models, dynamic charts, and AR / VR display technology to grasp the health status of the tunnel in real time, improving the efficiency and safety of monitoring work. ④ Improved decision support capabilities: By combining sensor data, machine learning algorithms, and BIM 3D models, the system provides more intelligent decision support. Maintenance personnel can scientifically formulate maintenance plans and emergency response strategies based on real-time assessment results.

[0134] Preferably, the risk assessment and early warning module aims to evaluate the health status of the tunnel structure in real time by setting thresholds and historical data models, and to dynamically monitor the tunnel based on the assessment results. This module can identify potential risks to the tunnel and, through a zoned monitoring and early warning mechanism, ensure timely alerts are issued when structural anomalies occur, prompting relevant personnel to take appropriate measures.

[0135] This module manages risk through the following core functions: Risk assessment model: Based on historical and real-time data, a risk assessment model is established to dynamically evaluate the structural health status of the tunnel and generate risk levels (such as normal, warning, danger, etc.). Risk zoning and monitoring: Risk zones are established based on different areas of the tunnel (such as entrances, bends, and joints), and independent monitoring standards and early warning mechanisms are set for each area. Early warning mechanism: When the monitored data exceeds the set threshold, the early warning system is automatically triggered to send alarm information to maintenance personnel or relevant personnel and update the risk area in real time so that timely and effective response measures can be taken.

[0136] Specific Implementation Methods: 1. Risk Assessment Model: Risk assessment is a dynamic health assessment of the tunnel by combining historical data and real-time monitoring data. The model is constructed based on the following elements: ① Historical Data: By analyzing long-term monitoring data of the tunnel, reference values ​​for each monitoring point under normal conditions are identified, and a historical data model is established. By learning from this historical data, the assessment model can identify the normal range of structural changes and potential risk areas. ② Real-time Data: Combining currently collected sensor data, the health status assessment is updated in real time. When there is a significant deviation between the real-time data and the historical data model, the model automatically calculates the risk value and makes a level judgment.

[0137] The risk assessment model determines the health status of a tunnel by setting multiple risk thresholds and analyzing trends in sensor data. These thresholds can be dynamically adjusted based on practical engineering experience and data analysis. For example, if the stress value of the tunnel wall exceeds a certain threshold, it indicates that there may be a risk of structural damage or instability in that area.

[0138] Risk assessment results: ;

[0139] in, Indicates at a point in time The results of the risk assessment of the tunnel, To monitor data in real time, For historical data models, The preset risk assessment threshold, This is the evaluation function. Based on this formula, the model can calculate the risk value and determine the risk level according to the deviation of real-time data.

[0140] The risk level is determined using a tiered model, which typically includes the following levels: normal: When the monitoring data is within a safe range, the tunnel is in normal condition. warn: When certain monitoring data approach or slightly exceed the preset safety threshold, the system issues a warning, indicating that there may be a problem with the health status of the tunnel. Danger: When monitoring data significantly exceeds safe limits, it indicates a major structural risk in the tunnel, requiring immediate emergency measures.

[0141] 2. Risk Zoning and Monitoring: Tunnel structures have different functional areas, and their safety requirements vary depending on the area. Therefore, this module divides the tunnel into several risk zones based on its structural characteristics and performs independent risk assessment and monitoring for each zone. The risk zoning implementation method is as follows:

[0142] Zoning Criteria: Based on the tunnel's structural characteristics and usage environment, it is mainly divided into the following zones: Tunnel entrance With exports: This area is where vehicles and people enter and exit, and it bears a heavy load, so it needs to be closely monitored. Curving area: Because curves can exert additional stress on the tunnel structure, curve sections need to be separately identified and closely monitored. Seam area: The junctions between different sections of a tunnel are often critical areas where cracks or stress concentrations occur, and should be given special attention. Base plate and top: The tunnel's floor and roof are subjected to pressure from above or groundwater, and may deform after long-term use. Zone warning: Within each area, the system sets independent risk thresholds and determines the risk level of that area based on changes in real-time data. This risk zoning allows for refined management of the tunnel, ensuring comprehensive monitoring of the safety of each area.

[0143] Risk Map: Based on monitoring data from different zones, the system generates a risk map of the tunnel, displaying the real-time health status and risk level of each area. The risk map visually presents the tunnel's safety situation using colors, graphics, and other methods, facilitating on-site monitoring by maintenance personnel.

[0144] Risk level: ;in, The risk level of a certain risk area. This is real-time monitoring data for the area. This is the safety threshold for the area.

[0145] 3. Early Warning Mechanism: When the monitored data exceeds the set threshold, this module can automatically trigger an early warning and take the following measures: Automatic alarm: When the system detects that data exceeds the safety threshold, it automatically generates an alert and sends it to relevant personnel (such as operations and maintenance personnel and administrators). The alert includes the risk area, specific issues, and recommended measures to help operations and maintenance personnel take action as quickly as possible. Real-time updates on risk areas: When monitoring data for a certain area shows an anomaly, the system will update the risk level of that area in real time and highlight it on the risk map so that staff can be aware of changes in tunnel risk in a timely manner. Emergency Response: The system works in conjunction with other management systems in the tunnel to automatically activate emergency response mechanisms when major risks occur, such as increasing monitoring frequency, initiating repair plans, and restricting passage. Warning status: ;in, Indicates the warning status (1 indicates a warning has been triggered, 0 indicates a warning has not been triggered). For current monitoring data, This is a preset safety threshold.

[0146] In this embodiment, the risk assessment and early warning module has the following technical effects: ①Accurate risk assessment and classification: Through a dynamic risk assessment model, this module can accurately evaluate the health status of the tunnel structure based on real-time monitoring data and historical data, and conduct zonal management according to the risk level to ensure the full safety of each area. ②Real-time early warning and rapid response: Once the monitoring data of the tunnel exceeds the preset threshold, the system can immediately trigger an early warning to ensure that the operation and maintenance personnel can discover problems in time and take emergency measures, thus avoiding the occurrence of major risks. ③Zonal management and refined control: By conducting risk zoning on the tunnel, the system can conduct refined monitoring of different areas. Risk zoning not only improves the accuracy of tunnel safety monitoring but also enables targeted maintenance and management according to the actual risk level. ④Decision-making support and safety guarantee: Based on the risk assessment and early warning mechanism of real-time data, this module can provide powerful decision-making support for tunnel management, improving the safety and operation efficiency of the tunnel structure.

[0147] Preferably, the event response and linkage module is mainly in the tunnel structure monitoring system. When the monitoring system has an anomaly or gives an early warning, it automatically triggers a series of response measures to ensure the safety of the tunnel. This module is not only responsible for the automatic initiation of emergency responses but also coordinates other management systems of the tunnel (such as the drainage system, lighting system, etc.) for dynamic adjustment and linkage control, thereby effectively reducing potential risks and losses. Through this efficient emergency response and linkage mechanism, this module ensures the timeliness and comprehensiveness of structural health management during the operation of the tunnel.

[0148] Specific functions include: ①Emergency response mechanism: When the monitoring data exceeds the set safety threshold or the system detects potential risks, the system automatically initiates emergency response measures, such as increasing the monitoring frequency, notifying relevant personnel, and starting repair plans. ②Linkage control: This module adjusts according to the risk assessment results through linkage with other management systems in the tunnel, such as drainage, lighting, ventilation, etc., and automatically takes corresponding protection or repair measures.

[0149] The implementation method of the event response and linkage module is as follows:

[0150] 1. Emergency response mechanism. The emergency response mechanism is an important part to ensure that the system can quickly take measures in case of a major early warning. This module decides whether to initiate emergency response measures based on the preset risk assessment threshold and real-time monitoring data. Monitoring frequency adjustment: When the system detects an anomaly or an increase in risk, it automatically increases the monitoring frequency. For example, if the stress or displacement change in a certain part of the tunnel exceeds the preset safety threshold, the system will automatically increase the monitoring frequency of this area to timely capture subtle changes and conduct accurate analysis.

[0151] Adjusted monitoring frequency: ;in, The adjusted monitoring frequency, Based on the base frequency, As a result of the risk assessment, if Then increase the monitoring frequency.

[0152] Maintenance personnel notification: When an alert is triggered, the system will send an alert message to designated operations and maintenance personnel via SMS, email, or mobile application. The alert content includes current monitoring data, risk areas, and suggested emergency measures.

[0153] Start the repair process: In some high-risk areas, the system can automatically initiate repair procedures, such as reinforcing the tunnel structure with automated equipment or activating emergency repair devices (such as grouting and support).

[0154] Self-diagnosis and repair: In some cases, the system may support self-diagnosis and self-repair of the device. For example, if signal loss occurs during data transmission, the system will automatically detect and switch to a backup communication line to ensure the continuity of data transmission.

[0155] 2. Interlocking Control: Interlocking control refers to the module's ability to coordinate with other management systems within the tunnel (such as drainage, ventilation, and lighting systems). When the system identifies potential structural risks in the tunnel, it automatically adjusts various operational parameters of the tunnel based on real-time risk assessment results, thereby reducing the spread or exacerbation of risks. Drainage system Collaboration: When the tunnel is at risk of flooding (e.g., due to excessive rainfall or seepage), the system automatically adjusts the drainage system's operating mode, such as increasing the operating frequency of the drainage pumps and opening more drainage channels to quickly remove excess water. Adjusted drainage pump power: ;in, The adjusted power of the drainage pump Based on power, For the results of the flood risk assessment, if This will enhance drainage capacity. Ventilation system connection move: When the tunnel detects excessively high concentrations of harmful gases or excessively high temperatures, the system can automatically adjust the ventilation system's airflow speed or activate backup ventilation equipment to ensure that the air quality and temperature inside the tunnel are maintained within a safe range. Adjusted ventilation speed: ;in, The adjusted ventilation speed, Based on ventilation speed, Based on the air quality risk assessment results, if This increases the ventilation intensity. Lighting system linkage:In the event of a risk event, the system can automatically adjust the brightness of the tunnel lighting system or activate the emergency lighting system to ensure the safe evacuation of personnel in the tunnel. Adjusted lighting brightness: ;in, For the adjusted lighting brightness, Based on brightness, For the lighting risk assessment results, if Then turn on the emergency lighting. Traffic control system linkage: When a tunnel experiences a major structural risk (such as a partial tunnel collapse or cracks), the system can automatically close the tunnel entrance or set up warning signs in front of the risk area to prevent vehicles or personnel from entering the dangerous area.

[0156] 3. Scalability and Flexibility of the Linkage Mechanism: The linkage control system is designed with a highly scalable and flexible structure, capable of connecting to various subsystems and dynamically adjusting according to the actual needs of different tunnels. For example, in addition to drainage, ventilation, and lighting systems, the system can also link with new monitoring systems based on the specific needs of the tunnel, further improving the level of intelligence in tunnel management. The system also features modular interfaces, allowing for seamless integration of new equipment and systems. Through open interfaces, users can add more intelligent devices in the future, such as surveillance cameras and alarm systems, according to the tunnel's operational needs.

[0157] 4. Risk Response and Data Updates: Upon triggering an emergency response, the system synchronously updates relevant data to the central processing platform and records the entire event response process. This data not only provides a basis for post-event analysis but also helps optimize future emergency response procedures. For example, by recording the time, measures, and effects of each emergency response, the system can learn and optimize its response strategies, improving the efficiency of future emergency handling.

[0158] In this embodiment, the event response and linkage module has the following technical effects: ① Highly efficient emergency response: Through an automated emergency response mechanism, it can quickly and accurately respond to various risks occurring in the tunnel, reducing the delay of manual intervention and improving processing efficiency. ② Intelligent linkage control: Linkage control with other systems in the tunnel ensures that the environment and structure within the tunnel are protected in a timely manner when a risk occurs, minimizing the spread of potential risks. ③ Enhanced safety: Linkage control is not limited to the response of a single system, but effectively improves the overall safety of the tunnel through multi-system linkage. By dynamically adjusting the working status of each system, it ensures that all equipment in the tunnel can maintain optimal operating status under abnormal conditions. ④ Flexible system expansion: This module supports the access of various devices and systems, ensuring that the tunnel management system can be flexibly expanded as needs change, adapting to future technological developments and operational requirements.

[0159] Example 2: Reference Figure 10-12 The tunnel structure monitoring and risk visualization early warning method of the present invention includes the following steps: S01, BIM model construction: obtain the BIM three-dimensional model of the tunnel through the tunnel design stage, and update the model in combination with the construction stage data to ensure its accuracy and completeness.

[0160] S02, Sensor Deployment and Data Acquisition: Various sensors are deployed at key structural locations in the tunnel to monitor the tunnel's status data (such as stress, strain, temperature, humidity, displacement, etc.) in real time.

[0161] S03, Data transmission and storage: The collected sensor data is transmitted to the data center for storage via wireless or wired networks and associated with the BIM model.

[0162] S04, Data Fusion and Intelligent Analysis: This involves fusing and processing the collected multi-source data, including noise reduction, standardization, and missing data completion. Machine learning or deep learning models are then used to analyze the data and generate a dynamic health status of the tunnel structure.

[0163] S05, Real-time Risk Assessment and Zoned Early Warning: Based on fused data, the risk level of the tunnel is assessed through algorithms, risk zones are formed, and corresponding early warning thresholds are set.

[0164] S06, Event Response and Linkage: When a risk exceeding the standard is detected, the system automatically triggers corresponding response measures, such as adjusting the monitoring frequency, activating the alarm system, or automatically performing repair measures.

[0165] As a preferred option: S01, BIM model construction and updating. During the tunnel design phase, a three-dimensional tunnel structural model is constructed using Building Information Modeling (BIM) technology. By combining data collected during the construction phase, the BIM model is updated in real time to ensure its synchronization and accuracy with the tunnel structural status. The BIM model is not only used to visualize the physical spatial layout of the tunnel, but also serves as the basic framework for monitoring data analysis.

[0166] Update Method: Structural monitoring data (such as displacement, temperature, and humidity changes) is continuously collected during the construction phase and compared with the original BIM model to update the model's geometry and material properties. The update algorithm uses an incremental approach, employing the following formula to update the BIM model: ;in, For a moment BIM model, This is an increment updated based on the latest data.

[0167] As a preferred option: S02, sensor deployment and data acquisition, various sensors are deployed at key structural parts of the tunnel (such as tunnel lining, support structure, connection nodes, etc.), mainly including stress sensors, strain sensors, displacement sensors, temperature and humidity sensors, and acceleration sensors. All sensors transmit the collected raw data to the data center in real time via wireless or wired networks to ensure the real-time performance and integrity of the data.

[0168] Deployment strategy: The sensor deployment was optimized based on the risk level of the tunnel structure and the characteristics of the construction process. A layered deployment method was adopted, distributing sensors in different monitoring sections to ensure data collection density in key areas.

[0169] Preferably, in step S03, data transmission and storage, the collected sensor data is transmitted to a data center for storage via wireless or wired networks and stored in association with the BIM model. Data compression and encryption algorithms are used during transmission to ensure data security and efficiency.

[0170] Data transmission and storage procedures: The collected data undergoes preliminary processing (such as data denoising and format conversion) at edge computing nodes, and is then transmitted to the central server through a high-bandwidth communication channel.

[0171] The data storage structure adopts a distributed storage system to ensure efficient storage and retrieval of large-scale data.

[0172] As a preferred option, S04, data fusion and intelligent analysis, involves preprocessing the collected multi-source data, including noise reduction, standardization, and missing value imputation, to ensure data quality. Various data fusion methods, such as weighted average, Kalman filtering, and particle filtering, are employed to fuse data from different sensors into a unified structural health assessment index.

[0173] Intelligent analysis methods: Deep learning-based time-series data modeling methods, such as Long Short-Term Memory (LSTM) networks, are used to predict the health status of tunnel structures. The LSTM model is trained on multiple time-series data to capture nonlinear and time-varying features. A model combining Convolutional Neural Networks (CNNs) and LSTMs is used for feature extraction and time-series prediction. This model can identify potential risk patterns in sensor data and perform health assessments.

[0174] Model formula: Assume the sensor dataset is Then, the health prediction results of the tunnel structure can be obtained by training the LSTM model: ;in, For predicted health status.

[0175] As a preferred option, S05, real-time risk assessment and zoned early warning, uses machine learning algorithms (such as Support Vector Machine (SVM), Random Forest (RF), XGBoost, etc.) to evaluate the risk level of the tunnel structure based on the fused data. Each risk zone corresponds to a different early warning threshold, ensuring accurate monitoring of every critical area of ​​the tunnel.

[0176] Risk assessment algorithm: Feature extraction and training are performed on the data for each region to form a risk assessment model for each region. By calculating the deviation between current data and historical data, it is determined whether the preset risk threshold is exceeded.

[0177] Risk level assessment formula: ;in, Risk level, For sensor data, This is a machine learning-based evaluation function.

[0178] As a preferred option, S06, event response and linkage, when the monitoring system detects that the risk level of a certain area exceeds a predetermined threshold, the system will automatically trigger a series of response measures. These measures include adjusting the monitoring frequency, activating the alarm system, dispatching repair equipment, or changing the construction plan.

[0179] Linkage mechanism: The system sets different response strategies based on different risk levels. Specific response measures are executed through an automated control system to ensure rapid action to mitigate risk.

[0180] Response strategy: ① Low risk: Only adjust the monitoring frequency and issue a warning. ② Medium risk: Activate the early warning system to remind the construction team to pay attention to safety. ③ High risk: Automatically start repair equipment or adjust the construction plan.

[0181] formula: ;

[0182] in, and These are risk level thresholds, each corresponding to a different response measure.

[0183] Furthermore, S03, Data Transmission and Storage, is a core component of the tunnel structure monitoring and risk early warning system. It involves the efficient and secure data flow from sensors to the data center. To ensure the system's efficiency, reliability, and security, this section is further detailed as follows:

[0184] The data transmission process involves the transmission links from various sensors to the data center. To achieve low-latency, stable communication and ensure the real-time performance of large-scale sensor data, the following steps and technologies are employed:

[0185] Preliminary processing by edge computing nodes: After sensor data acquisition, the data is first processed by edge computing nodes. This step includes noise reduction, missing data imputation, and data format conversion to improve data quality.

[0186] Kalman filtering and other algorithms are used to denoise the sensor data, reducing the impact of environmental interference on the sensor data.

[0187] The collected data is synchronized in time and standardized in format, and converted into a unified data format that is compatible with BIM models and other systems. Kalman filter algorithm: ;in, This is an estimated value. The sensor measurement value. Here is the gain matrix. This is the measurement matrix. Transmission Channel and Protocol Discussion: Choose a high-bandwidth wireless communication channel (such as 5G, Wi-Fi 6) or wired fiber optic network to ensure low latency and high bandwidth requirements during data transmission. Data transmission uses Transmission Control Protocol (TCP) or User Datagram Protocol (UDP), selecting the appropriate protocol based on real-time and reliability requirements.

[0188] To ensure data transmission security, encryption algorithms (such as AES-256) are used to encrypt the data, preventing leakage or tampering during transmission. AES-256 encryption algorithm: ;in, Plain text data For encryption key, This is encrypted data.

[0189] To handle large amounts of sensor data and provide efficient data query capabilities, the system employs a distributed storage architecture for data storage. The specific implementation details are as follows: Distributed storage system: Distributed databases (such as HDFS and Ceph) are used to store the transmitted data, ensuring that massive amounts of data can be accessed on demand. Distributed storage systems can automatically scale to adapt to the needs of monitoring projects of different sizes. Data redundancy: To avoid single points of failure, data redundancy technologies (such as RAID and Erasure Coding) are used to ensure high availability and fault tolerance of data during storage. Erasure Coding Storage principle: The data is divided into multiple blocks, with some blocks stored as redundant information to ensure data recovery in the event of hardware failure. Storage associated with BIM model:Sensor data is linked and stored with the BIM model, forming a two-way link between data and model. Metadata stored in the database (such as data acquisition time, location, sensor number, etc.) establishes a connection between sensor data and corresponding tunnel structural components.

[0190] The database design uses graph databases (such as Neo4j) for modeling, which can effectively handle complex spatial relationships and multi-level data associations, ensuring the efficiency of data querying.

[0191] Graph database query relationships: ;

[0192] Query the status data of a tunnel area monitored by a specific sensor.

[0193] Storage security and backup: All data is encrypted during storage to ensure its security. Strong encryption algorithms (such as RSA and AES) are used to ensure data is not accessed illegally during storage. Regular incremental and full backups are performed, and off-site backups are used to ensure data durability and prevent data loss due to disasters or hardware failures. Backup strategy: Combining incremental backup and full backup ensures data integrity and backup efficiency.

[0194] The core of this method lies in the data transmission and storage process. It employs efficient edge computing nodes, stable communication protocols, encryption technology, and a distributed storage architecture to ensure the efficient and secure transmission and long-term storage of tunnel monitoring data. Simultaneously, through association storage with the BIM model, sensor data and the tunnel structure model can be dynamically updated and linked for querying, improving the accuracy and real-time performance of data analysis and decision-making.

[0195] Furthermore, S04, Data Fusion and Intelligent Analysis, plays a crucial role in tunnel structure monitoring. Its purpose is to integrate multi-source, heterogeneous data into unified structural health assessment indicators and conduct intelligent analysis to predict the health status and risks of the structure. The following is a further refinement of this process.

[0196] 1) Data fusion: The goal of data fusion is to unify data from different sensors into standardized health assessment indicators for subsequent analysis. This process involves multiple steps and methods:

[0197] (1) Data preprocessing: Noise reduction: Noise in sensor data is removed by algorithms such as Kalman filtering or particle filtering to ensure data accuracy. Kalman filtering: This method is suitable for estimating the state of a system in the presence of noise. Assume the state of the system is given by the following equation: ; ;in, This is the current state. Here is the state transition matrix. To control the input matrix, For process noise, For measured values, For measuring noise. Missing value supplementation: Interpolation algorithms (such as linear interpolation, spline interpolation, or data imputation) are used to fill in missing values ​​in sensor data to ensure data continuity. For time series data, interpolation algorithms can be used for data imputation. ; standardization: To avoid the influence of differences in the dimensions of different sensors, all data are standardized to have a mean of 0 and a variance of 1. The standardization formula is: ;in, The mean of the data. The standard deviation is denoted as .

[0198] (2) Fusion method: Weighted average method: Data from different sensors is fused by assigning different weights to it. The resulting health assessment indicators are then analyzed. The weighted average is obtained as follows: ;in, For the first Data weights of each sensor, Data collected by the sensor. Kalman filtering: By iterating through sensor data multiple times, a mathematical model is used to make an optimal estimate of the system state, reducing errors in the data and fusing them together. Particle filtering: Suitable for nonlinear, non-Gaussian noise environments, it uses a set of particles for Monte Carlo estimation and fuses data from different sensors to estimate the structural health status.

[0199] 2) Intelligent analysis methods: These methods utilize deep learning models to predict structural health status and assess risks. The following is a further refinement of the two main methods:

[0200] (1) LSTM-based time series data modeling: Long Short-Term Memory (LSTM) networks are an effective tool for processing time series data, especially suitable for capturing nonlinear and time-varying features in the data. In tunnel structure monitoring, LSTM can predict the future structural health status by learning the temporal correlation in historical data.

[0201] LSTM Model Structure: LSTM controls the information flow through multiple gates (input gate, forget gate, and output gate), making it suitable for modeling long-term series data. The mathematical formula for LSTM is as follows:

[0202] ;

[0203] ;

[0204] ;

[0205] ;

[0206] ;

[0207] in, , , These are the forget gate, input gate, and output gate, respectively. In cellular state, It is in a hidden state.

[0208] Health status prediction: Assuming the sensor dataset is Then the output of the LSTM model is:

[0209] ;

[0210] in, The health status predicted by the model reflects the risk level of the tunnel structure at future moments.

[0211] (2) Feature extraction and temporal prediction using CNN and LSTM combination: Convolutional Neural Networks (CNN) are good at extracting local features from data, while LSTM is good at processing temporal data. Combining CNN and LSTM can better extract spatial and temporal features from temporal data and enhance the model's predictive ability.

[0212] CNN Feature Extraction: CNNs extract local spatial features from input data through convolutional layers, which is highly effective for identifying latent patterns in sensor data. The formula for calculating a convolutional layer is: ;

[0213] in, For convolution kernel, For input data, For bias, This indicates a convolution operation.

[0214] This model combines CNN and LSTM: First, a CNN is used to extract spatial features from the sensor data. Then, the extracted features are input into an LSTM network for temporal prediction. This model structure can capture both spatial and temporal information in the data, thereby effectively identifying risk patterns.

[0215] Model formula: ;in, This represents the result after performing CNN feature extraction on the sensor data. This indicates that the extracted features will be input into the LSTM model for time series prediction;

[0216] Data fusion and intelligent analysis are core components of tunnel structural health assessment. Various data fusion techniques (such as weighted average, Kalman filtering, and particle filtering) are used to integrate data from different sensors into a unified structural health assessment index. Combining deep learning models such as LSTM and CNN effectively captures the temporal and spatial characteristics of the data, thereby enabling health prediction and risk assessment of tunnel structures.

[0217] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A tunnel structure monitoring and risk visualization early warning system based on digital twins, characterized in that, include: The sensor monitoring module deploys various sensors within the tunnel and collects data using a unified time scale. The data acquisition and transmission module buffers and re-transmits multi-source data at the edge, and then uploads it to the center via a secure link. The BIM 3D model module builds and maintains a tunnel BIM 3D digital twin model, stores the geometric, material and maintenance attributes of components, and binds monitoring points to model components in two directions and supports observation-based dynamic updates. The data fusion and analysis module performs denoising, interpolation, standardization, semantic alignment and fusion on multi-source data, and identifies health status and predicts trends based on a pre-trained deep learning model and combined with structural topology and working condition information. The risk assessment and early warning module assesses the risks of each zone based on the fusion results, sets multi-level trigger conditions, and generates disposal recommendations. The visualization module displays monitoring and risk information on the BIM 3D model through zone coloring, timeline playback, and component-level drill-down, and provides augmented reality on-site guidance to the terminal. The event response and linkage module implements parameterized linkage control of drainage, ventilation, lighting and alarm devices and records the closed-loop results when an early warning is triggered.

2. The system according to claim 1, characterized in that, The BIM 3D model module uses an incremental update method to write monitoring-driven geometric and attribute changes into the digital twin, and records version evolution with a unique component identifier and a time window. The increment satisfies: ； in For a moment The twin state, Establish mapping relationships with components, measurement points, and processing versions for playback and traceability.

3. The system according to claim 1, characterized in that, The data acquisition and transmission module performs interpolation and standardization preprocessing and records data quality labels at the edge side, and the steps satisfy: ， ； It employs caching and breakpoint retransmission to maintain integrity, and uses encrypted channels and verification to maintain security and consistency.

4. The system according to claim 1, characterized in that, The data fusion and analysis module deeply integrates multi-source sensing and BIM component attributes to form health indicators and constructs a data lineage diagram of "component-measuring point-time window-version"; the fusion and indicator calculation satisfy the following: ， ； in For the first Source data, The value of is constrained by confidence level and spatiotemporal correlation.

5. The system according to claim 1, characterized in that, The data fusion and analysis module suppresses drift and jump noise based on state-space recursion and observation correction, and performs consistency comparison with the deep learning output; the recursive relationship satisfies: ； and with ； The results are updated to obtain a corrected sequence of health indicators for subsequent zoning assessments.

6. The system according to claim 1, characterized in that, The risk assessment and early warning module sets multi-level thresholds and triggering strategies, providing alarm, early warning, and emergency response levels for components, zones, and sections respectively. When triggered, the event response and linkage module automatically adjusts the sampling frequency, drainage pump power, and ventilation speed according to the response table, and writes the response results and time-series trajectory back to the digital twin timeline. The triggering can be characterized by the following formula: ； in As a current risk measure, This corresponds to the threshold.

7. A method for tunnel structure monitoring and risk visualization early warning based on digital twins, characterized in that, include: S1, construct and initialize the BIM 3D digital twin model, and complete the two-way binding of monitoring points and model components and metadata registration; S2, convergent displacement sensors, strain gauges, accelerometers, tilt sensors, and piezometers are deployed at key locations such as the lining, arch waist, arch crown, invert, and joints, and data are collected using a unified time scale. After edge buffering and breakpoint retransmission, the data is uploaded. S3 performs denoising, interpolation, standardization, and semantic alignment, performs multi-source fusion, and outputs health status recognition and trend prediction based on a pre-trained deep learning model combined with structural topology. S4. Based on the fusion results, risk classification and multi-level trigger determination are carried out for each region to form disposal suggestions and linkage parameters; S5 performs zoning and coloring, timeline playback, and component-level drill-down display on the BIM 3D model and augmented reality interface. When the trigger conditions are met, it implements parametric linkage control of drainage, ventilation, lighting, and alarm devices, records the closed loop, and writes back the twin version.

8. The method according to claim 7, characterized in that, S3's deep learning prediction adopts a joint approach of spatial feature extraction and temporal modeling. First, the sensor and component features are encoded, and then temporal correlation modeling is performed to obtain the state estimate and uncertainty range of the future window. The training samples consist of historical monitoring data, structural inspection results, and maintenance records, and are updated in small batches online using new data.

9. The method according to claim 7, characterized in that, The time-series risk assessment of S4 employs a joint mapping of historical and real-time data and outputs partitioned quantification results, which can be characterized by the following formula: ， as well as: ； in For the current window data, For historical sample sets, These are the model parameters, and the results are used to determine the warning level and response priority.

10. The method according to any one of claims 7-9, characterized in that, The triggering and linkage of S5 are executed synchronously at the component level and the zone level: when the trigger condition is determined, the target component and adjacent components are highlighted in the BIM 3D model and the handling steps are pushed, the monitoring sampling frequency is automatically increased, the drainage pump power and ventilation speed are increased, and lighting and audible and visual alarms are linked when necessary; after the handling is completed, a handling report is generated, and it is written into the digital twin timeline according to the component identification and time window for playback and auditing.

Images