Digital twin system and method for intelligent construction, operation maintenance and emergency management of urban bridge

Abstract

The invention belongs to the field of bridge engineering, and provides a digital twin system and method for intelligent construction, operation maintenance and emergency management of an urban bridge. On the basis of fully collecting bridge engineering construction and operation maintenance data, the system comprises the full life cycle of bridge construction and operation maintenance. Bridge real-time structure state evaluation, damage identification, reliability evaluation, predictive maintenance and closed-loop control scientific decisions related to bridge intelligent monitoring, health monitoring, operation maintenance and safety emergency management based on ubiquitous perception and artificial intelligence deep learning algorithms are realized. BIM + GIS rapid and accurate modeling integration based on three-dimensional point cloud and related to bridge disease prediction and processing and a digital twin system, and AI Agent efficient collaboration based on an AI large model / medium model / small model. According to the invention, system support or method guidance can be provided for intelligent construction, operation and maintenance and emergency management of the large-span bridge.

Description

Technical Field

[0001] This invention relates to the field of intelligent construction and operation and maintenance technology in civil engineering, particularly to the field of construction, operation and maintenance monitoring of bridge engineering, and especially to application tools and implementation methods for intelligent health monitoring and risk management of bridges. Specifically, it is a digital twin system and method for intelligent construction, operation, maintenance and emergency management of urban bridges. Background Technology

[0002] At present, the level of informatization and intelligence in bridge construction and operation is still relatively limited. Although there is a relatively rich data type, it is possible to conduct simple structural assessment and analysis and prediction. However, it lacks bridge structure simulation and cannot conduct more complex bridge disease analysis and prediction, intelligent operation and maintenance decision-making, intelligent formulation of maintenance plans, optimization of maintenance funding plans, etc. With the rapid development of technology, bridge operation and maintenance management has been gradually improved. Technologies such as BIM, GIS, HIM, AIM, Internet of Things, artificial intelligence, deep learning, and big data have been integrated into it. The perception accuracy, data collection efficiency, digital model accuracy, digital model timeliness, and interaction level between physical and digital models of bridge operation and maintenance data have been greatly improved. (Hong Kong-Zhuhai-Macao Bridge Authority, Research on the Architecture of Intelligent Maintenance System for Cross-Sea Bridges Based on Digital Twin Concept: 0451-0712 (2023) 04-0383-09. 2023-4).

[0003] The bridge intelligent health monitoring and digital twin system realizes the accurate mapping of physical entities in digital space and achieves complex interaction between physical and digital models through data-driven methods. During the bridge operation and maintenance phase, the establishment of a bridge digital twin model can realize static and dynamic data management in the bridge operation and maintenance process, provide decision-making solutions for bridge operation and maintenance technicians, and realize intelligent and smart management of bridge operation and maintenance. The digital twin, combined with new surveying and mapping, identification perception, collaborative computing, full-element expression, simulation and deep learning, has initially formed a digital twin system with these nine cores. Digital twin technology will inevitably become the backbone of the future development of bridge operation and maintenance management models. (Hong Kong-Zhuhai-Macao Bridge Authority, Research on the Architecture of Intelligent Maintenance System for Cross-Sea Bridges Based on Digital Twin Concept: 0451-0712 (2023) 04-0383-09. 2023-4).

[0004] However, at present, there is a lack of mature digital twin systems for bridge engineering both domestically and internationally. Most existing digital twin systems simply digitize paper documents from bridge engineering projects or perform simple data summarization and classification. Because various data types operate independently, the correlation between them is poor, resulting in low data retrieval efficiency. In terms of functionality, existing digital twin systems only perform simple structural analysis of bridge structures and cannot perform more complex bridge defect analysis, structural condition prediction, or maintenance plan decisions. Summary of the Invention

[0005] To address at least one of the problems existing in the prior art, this invention provides a digital twin system and method for intelligent construction, operation, maintenance, and emergency management of urban bridges. This system enables comprehensive intelligent construction and operation of long-span bridges, allowing for bridge health monitoring, structural condition assessment, bridge defect prediction, and intelligent maintenance decision-making during the construction and operation process. This improves the accuracy, reliability, and timeliness of health monitoring for long-span bridges, while saving economic expenses.

[0006] A digital twin method for intelligent construction and emergency management of urban bridges includes the following steps: S01. Establish a three-dimensional point cloud building information model (BIM) and a geological information system (GIS) based on UAV lidar and photogrammetry technology.

[0007] S02. Establish a finite element calculation model based on refined BIM and GIS, perform structural mechanical stress verification and safety assessment of the bridge, and regularly correct the finite element model using collected monitoring data.

[0008] S03. Conduct loading tests based on a scaled-down model of the bridge to perform structural mechanics stress calculations and safety assessments of the bridge.

[0009] S04. Conduct bridge structural load tests, and perform structural mechanical stress calculations and safety assessments on the bridge.

[0010] S05. Through the following three methods: 1. Bridge shaking table test based on a scaled-down model of the bridge; 2. Finite element numerical simulation of bridge seismic response based on the finite element method; 3. Bridge forced vibration response test, establish a seismic response model for long-span bridges, analyze the vibration mechanism and predict the seismic response.

[0011] S06. The following three methods are used: 1. Wind tunnel experiments on bridges based on scaled-down models of bridges; 2. Finite element numerical simulation of bridge wind vibration response based on computational fluid dynamics (CFD); 3. Field measurements to establish a wind vibration response model for long-span bridges, analyze the wind vibration principle, and predict the wind vibration response.

[0012] S07. Integrate the stress calculation data, wind vibration response data, and seismic response data of long-span bridge structures obtained by model tests, numerical simulations, and field measurements. Through data-driven deep learning algorithms, establish the inherent logical relationship of the dataset and collect monitoring data of the operational status of long-span bridges in real time. Predict and evaluate the safety status of bridges in real time, thereby constructing an intelligent health monitoring system for long-span bridges.

[0013] S08. Based on the intelligent health monitoring system, construct an intelligent closed-loop control digital twin system that integrates intelligent monitoring, intelligent data transmission, intelligent data storage, intelligent data analysis, and intelligent structural disease prediction and treatment, so as to realize automatic perception, machine learning, intelligent prediction, scientific evaluation and automatic execution throughout the entire bridge operation process.

[0014] S09. Continuously collect monitoring data and construct a digital twin system for long-span bridges. By comparing the error between the calculated value and the measured value of the digital twin, the digital twin is continuously corrected, thereby realizing the construction and iterative optimization evolution of a digital twin system for intelligent construction, health monitoring, operation and maintenance and emergency safety management of urban long-span bridges.

[0015] Establish an intelligent monitoring system based on edge sensors and UAV payloads; a real-time structural status assessment, defect prediction, and operation and maintenance AI agent decision-making system for bridges based on deep learning and artificial intelligence algorithms. A digital twin system and method for intelligent construction, operation, maintenance, and emergency management of urban bridges includes a baseline modeling module, a ubiquitous sensing module, a virtual-real interaction module, a high-fidelity digital twin system (accurate prediction) module, a forward design module, a digital construction module, a full-state operation and maintenance module, and a closed-loop control module. The benchmark modeling module is used to build digital models, which include a full-state simulation model of the physical system, an AI-based reduced-order physical system model (AID-ROM), and a geometric model.

[0016] The AI-based physical system reduced-order model (AID-ROM) construction includes a finite element model for representing the stress characteristics of the bridge structure, a structural simulation model for representing the seismic response model and wind vibration response model of the bridge under seismic load and wind load, a CFD simulation model including hydrodynamic information model (HIM) and air flow field information model (AIM), and a geotechnical engineering ground motion response simulation model corresponding to the seismic wave field information model (EIM); the geometric model construction includes a building information model (BIM) for representing the bridge geometry and a geological information model (GIS) for representing the bridge hydrological environment and geological environment.

[0017] The benchmark modeling module employs modern digital modeling technologies such as spaceborne and airborne photogrammetry, and laser point cloud analysis to establish corresponding digital models. These include finite element models (FEM) representing the structural stress characteristics of bridges, building information models (BIM) of bridge geometry, geological information models (GIS) of the bridge's hydrological and geological environment, and dynamic models (FEM+CFD), hydrodynamic information models (HIM), airflow field models (AIM), and seismic wavefield information models (EIM) of bridges under seismic and wind loads to demonstrate their seismic and wind-induced vibration responses. When establishing a precise bridge point cloud model, a UAV equipped with optical cameras, thermal imaging cameras, lidar, and microwave radar can automatically plan the UAV's flight path using a pre-established bridge BIM, achieving precise UAV modeling of the bridge. The resulting high-precision point cloud model (millimeter-level) can be used to establish a refined three-dimensional solid model of the bridge through artificial intelligence, 3D reconstruction, and plane fitting technologies.

[0018] The ubiquitous sensing module is used to acquire stress and strain data of bridge structural components. By deploying corresponding edge devices, sensors, total stations, etc. on the bridge structural surface, an intelligent bridge monitoring system is established to monitor the bridge's environmental condition, operational load, and static and dynamic response. Simultaneously, satellites, UAVs, and interferometric synthetic aperture radar (InSAR) mounted on bridge monitoring vehicles can be used to achieve integrated, precise deformation monitoring of long-span bridges from the air, space, and ground. The measurable data includes bridge span center deflection and displacement.

[0019] The virtual-real interaction module, in the digital twin system for intelligent construction and emergency management of urban bridges, is embodied in the digital mapping and two-way interaction function between the digital twin and the physical entity. 5G mobile communication enables rapid information transmission between the digital twin and the actual bridge, as well as the full-scale (scaled-down) model of the bridge. Through the two-way transmission and interaction of data predicted by the digital twin system with sensor data from the actual bridge and the full-scale (scaled-down) model, the stress state of the bridge can be predicted in real time, ensuring construction safety.

[0020] The high-fidelity digital twin system (accurate prediction) module is used to characterize and describe the geometric, physical, temporal, behavioral, rule-based, evolutionary, evolutionary, update, redundancy, failure, and uncertainty quantification aspects of the benchmark modeling module and the ubiquitous sensing module from multiple dimensions of constitutive relations, multiple time scales, multiple spatial scales, and material-level, component-level, part-level, system-level, and system-level construction, as well as the system update methods after reinforcement and maintenance. The high-fidelity digital twin system (accurate prediction) module includes a reduced-order structural simulation model from the model-driven approach, and data from the data-driven approach, including a series of geometric, physical, temporal, behavioral, loading and unloading history information, dimensionality reduction calculation schemes, and result output definitions. By integrating model data from the model-driven and data-driven approaches, the high-fidelity digital twin system (accurate prediction) module constructs representative features of the learning data across multiple processing layers using dynamic graph neural networks and graph deep learning methods. It explores complex patterns and correlations between input and output data, and performs real-time calibration and correction of the digital twin model using intelligent monitoring data, thereby achieving the assessment and prediction of the bridge structure's health status. Meanwhile, the high-fidelity digital twin system (accurate prediction) module can use the latest data from the actual operation and maintenance process of long-span bridges collected in real time by UAV photogrammetry, laser scanning, sensors and other equipment to calibrate and correct the digital twin model. Through model iteration, the accuracy and prediction accuracy of the digital twin model are continuously improved to achieve the goal of accurate prediction.

[0021] The forward design module involves monitoring the bridge's stress state through a ubiquitous sensing module during bridge construction using a pre-established digital twin (FEM+BIM+GIS, etc.), and updating the digital twin in stages as construction progresses. This dynamically changing digital twin can guide on-site bridge construction through real-time predictions and monitor the bridge's stress characteristics in real time, ensuring safety during construction. The forward design module addresses quality control issues in the actual construction of long-span bridges. This module uses the baseline modeling module within the digital twin system to convert the point cloud model of the steel / concrete box girder into a three-dimensional solid model, simulating the pre-assembly of the steel / concrete box girder in the virtual environment of the digital twin system. This enables intelligent construction of long-span urban bridges through virtual-real interaction and construction between the digital twin and the physical entity.

[0022] The digital construction module is the process of automatically operating digital construction machinery by controlling the digital twin system after the forward design module is completed. Practical applications include: (1) using the digital twin system to automatically cast concrete beam segments and automatically hoist steel box girders using intelligent unmanned hanging baskets and unmanned cranes; (2) using intelligent prestressed tensioning equipment to automatically tension prestressed steel bars and anchor the tensioning ends; (3) using intelligent winches to automatically tension cable stays; and (4) using the digital twin system to command steel mesh robots, concrete pouring robots, and concrete curing robots to perform a series of automated construction steps such as steel mesh binding, concrete pouring, and concrete watering curing during the secondary paving of the bridge deck.

[0023] The comprehensive operation and maintenance module within the bridge digital twin system encompasses multi-dimensional information on the entire modality and all states of bridge construction and operation. The comprehensive modality includes bridge-related information carriers such as sound, light, electricity, text, images, and video, while the all states represent the spatiotemporal stages of the bridge structure. By integrating this comprehensive bridge information, the digital twin system can perform predictive maintenance and intelligent management throughout the entire lifecycle of bridge construction and operation.

[0024] The closed-loop control module provides real-time risk management strategies for the construction and operation phases of long-span bridges. Monitoring data collected by edge devices and sensors deployed on the bridge surface is uploaded via the internet to a high-fidelity digital twin system (precise prediction) module on the server for evaluation and prediction. The high-fidelity digital twin system (precise prediction) module also includes an artificial intelligence agent (AI Agent). The AI ​​Agent is an application software package driven by a pre-trained large AI model. The pre-training set includes predicted stress characteristics calculated from material parameters in forward analysis, and material parameters derived from on-site monitored stress characteristics in inverse analysis. Through pre-training, the AI ​​Agent can derive sub-products during the iteration process, namely, large AI models (hundreds of billions of neurons), medium AI models (hundreds to tens of billions of neurons), and small AI models (hundreds of thousands to tens of thousands of neurons). Based on this, various AI models can automatically analyze the results of the above assessment and prediction, combined with different specific application scenarios (specific parameters), and provide corresponding suggestions for risk management of long-span bridges to relevant industry personnel, thereby achieving the goal of intelligent decision-making. The closed-loop control module also has functions for safety risk control, emergency management, and automatic execution. For example, it can automatically issue warnings or urgently evacuate people and vehicles when the wind speed in the flow field around the bridge is detected by the ubiquitous sensing module and the bridge's wind vibration exceeds the standard, as predicted by the accurate prediction module; it can automatically restrict traffic flow through traffic lights when the bridge deck is overloaded; and it can notify maintenance personnel to carry out repairs and reinforcement when a drone detects cracks in the bridge deck. The closed-loop control module needs to intelligently make decisions and regulate the behavior and operating status of personnel, machines, materials, methods, and the environment during the construction, operation, and maintenance of long-span bridges, achieving the goals of real-time early warning, real-time feedback, and real-time control.

[0025] Preferably, the baseline modeling module includes the construction of an AI-based physical system reduced-order model (AID-ROM) and the construction of a geometric model; Among them, the AI-based physical system reduced-order model (AID-ROM) constructs structural simulation models including finite element models, seismic response models and wind vibration response models, as well as CFD simulation models including hydrodynamic information models (HIM) and air flow field information models (AIM). The geometric model construction includes Building Information Modeling (BIM) and Geological Information Modeling (GIS).

[0026] Preferably, the ubiquitous sensing module includes an environmental condition monitoring module, an operational load monitoring module, and a static and dynamic response monitoring module; Environmental condition monitoring includes temperature, wind direction, customs, bridge pier erosion, and steel corrosion monitoring; Operational load monitoring includes highway vehicle load, lane load, and pedestrian load monitoring; The static and dynamic response monitoring includes the overall structural geometric deformation, stress distribution and changes in local components, and cable force monitoring of the stay cables.

[0027] Preferably, the virtual-real interaction module includes digital mapping and two-way interaction between the bridge digital twin and the actual bridge, and between the full-scale (scaled-down) model.

[0028] Preferably, the high-fidelity digital twin system (accurate prediction) module includes a digital twin model calibration and correction module, a digital twin model module, a structural status assessment module, and a bridge defect prediction module; The technologies used include cloud computing, blockchain, big data, Internet of Things, photogrammetry, laser scanning, large AI models, model reduction, virtual simulation, machine learning, computer vision, deep learning, artificial intelligence (AI), edge devices, augmented reality (AR), and virtual reality (VR).

[0029] Preferably, the forward design module includes bridge stress state monitoring and model iterative updates.

[0030] Preferably, the digital construction module includes a bridge digital twin system controlling unmanned hanging baskets, unmanned cranes, intelligent prestressing tensioning equipment, intelligent winches, steel mesh robots, concrete pouring robots, and concrete curing robots for automated construction.

[0031] Preferably, the full-state operation and maintenance module includes full-modal information in the form of information carriers such as bridge sound, light, electricity, text, images, and video, as well as full-state information of the spatiotemporal stage of the bridge structure.

[0032] Preferably, the closed-loop control module uses an artificial intelligence agent (AI Agent) trained by dynamic graph neural network and graph deep learning method to propose bridge risk management suggestions and countermeasures, and to carry out safety risk control and emergency management. Practical applications include: (1) real-time warning when potential risks of the bridge are detected, and emergency evacuation of people and vehicles; (2) real-time feedback to facilities such as bridge traffic lights to control the traffic flow on the bridge; and (3) real-time feedback to the bridge operation and maintenance department so that bridge engineers can organize personnel to carry out bridge reinforcement and maintenance work in a timely manner.

[0033] Compared with the prior art, the beneficial effects of the present invention are at least as follows: This invention provides a digital twin system and method for intelligent construction, operation, maintenance, and emergency management of urban bridges, comprising eight modules: baseline modeling, ubiquitous sensing, virtual-real interaction, accurate prediction, forward design, digital construction, full-state operation and maintenance, and closed-loop control. It enables comprehensive intelligent construction and operation and maintenance of long-span bridges, achieving bridge health monitoring, structural status assessment, bridge defect prediction, and intelligent maintenance decision-making during the construction and operation and maintenance process. This improves the accuracy, reliability, and timeliness of health monitoring for long-span bridges, while saving economic expenses.

[0034] In past engineering projects, the lack of integrated software for analyzing various data, whether from bridge construction monitoring or operational monitoring, has resulted in engineers having to use different software or methods to analyze and verify different monitoring data, leading to low efficiency. The high-fidelity digital twin system (accurate prediction) module in this invention includes constitutive relationships for various finite element models. It can perform inversion analysis using data such as strain, displacement, and wind speed from the monitoring data to verify and evaluate the stress state of the overall bridge structure. Due to the AI-based AID-ROM model reduction and the construction of logical relationships (hidden layers) between input and output data using dynamic graph neural networks and graph deep learning methods, the model prediction speed will be significantly faster than traditional finite element analysis models.

[0035] Because various specialized software programs (such as drawing software, finite element analysis software, 3D modeling software, point cloud processing software, construction progress management software, etc.) often lack interfaces that support each other's data formats, it is difficult to transfer data output from different software programs to each other, thus reducing the efficiency of data analysis. On the other hand, data and materials in various carriers (such as design drawings, bridge monitoring data, digital models, construction progress plans, etc.) usually lack a comprehensive management system with high visualization, convenience, efficiency, and ease of use to facilitate data retrieval or collaboration among construction personnel in various departments. The large-span bridge digital twin system constructed in this invention includes a front-end and a back-end. The back-end can perform mutual transfer and conversion between various models and data through secondary development, thereby reducing the workload of pre-processing and post-processing during data analysis and greatly enhancing work efficiency. The front-end, through the design of a highly visualized digital twin platform, facilitates engineers to monitor the stress state of the bridge structure in real time during the construction and operation phases. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of the digital twin system for intelligent construction and emergency management of urban bridges in an embodiment of the present invention.

[0037] Figure 2 This is a schematic diagram of the composition of the ubiquitous sensing module in an embodiment of the present invention.

[0038] Figure 3This is a schematic diagram of the modular structure of the digital twin system for intelligent construction and emergency management of urban bridges in an embodiment of the present invention. Detailed Implementation

[0039] Exemplary embodiments of the present invention will now be described in more detail with reference to the accompanying drawings. However, it should be understood that the invention can be implemented in various forms and should not be limited to the exemplary embodiments set forth herein; rather, these embodiments are provided to enable a more thorough understanding of the invention and to fully convey the scope of the invention to those skilled in the art. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0040] like Figure 1 As shown in the figure, the digital twin method for intelligent construction and emergency management of urban bridges provided by this invention includes the following steps in the intelligent construction and operation process: S01. Establish a three-dimensional point cloud building information model (BIM) and a geological information system (GIS) based on UAV lidar and photogrammetry technology.

[0041] In one embodiment of the present invention, in specific implementation, novel surveying, mapping, BIM, GIS modeling technologies, LiDAR (Light Detection and Ranging) technology, photogrammetry technology, and UAV technology are combined. Original 3D point cloud data of long-span bridges is collected based on LiDAR technology. After preprocessing such as noise reduction and downsampling, point cloud models are established by using automatic point cloud registration (orientation) technology and adjustment methods such as least squares to reduce errors. In other embodiments, photogrammetry technology can be used to establish point cloud models by stitching together photos with a certain degree of overlap collected by an RTK UAV through oblique photography. Based on the point cloud models obtained by LiDAR or photogrammetry, a 3D solid model is further generated using a point cloud-to-2D graphics algorithm (or artificial intelligence algorithm), ultimately establishing a refined BIM and GIS model.

[0042] Compared to traditional BIM and GIS modeling, the technology of generating 3D solid models from point clouds has high precision and high timeliness, enabling the digital twin system to perform functions such as real-time perception, real-time modeling, real-time correction, and real-time prediction.

[0043] S02. Establish a finite element calculation model based on refined BIM and GIS, perform structural mechanical stress verification and safety assessment of the bridge, and regularly correct the finite element calculation model using collected monitoring data (point cloud models collected at different construction stages).

[0044] In one embodiment of the present invention, in specific implementation, BIM and GIS refined modeling technology are combined. Refined BIM and GIS are transformed into finite element model algorithms, finite element simulation methods, and finite element model correction methods to establish a finite element calculation model for urban long-span bridges.

[0045] High-precision BIM and GIS are used to build finite element models and perform finite element analysis. Compared to traditional finite element modeling, the finite element calculation model generated by this method has higher accuracy, higher intelligence, and higher timeliness, conforming to the self-perception, self-inference, self-learning, self-evaluation, self-prediction, and self-execution characteristics of this digital twin system. Traditional finite element modeling is often subject to many limitations when revising finite element models, such as the inability to accurately reflect the actual condition of long-span bridges in the existing finite element model. The method of converting finite element models using high-precision BIM and GIS can effectively avoid this situation. High-precision BIM and GIS generated from point cloud models collected by LiDAR and photogrammetry technologies can collect and quickly reflect the latest condition of long-span bridges in real time, facilitating accurate and rapid finite element model revision.

[0046] S03. Conduct loading tests based on scaled-down / full-scale models of the bridge's overall / partial components to perform structural mechanics stress verification and safety assessment of the bridge.

[0047] In one embodiment of the present invention, a loading test based on a scaled-down model of the bridge is conducted to perform structural mechanics stress calculations and safety assessments of the bridge. Specifically, in conjunction with the bridge structural model loading test, a bridge structural model is established that is geometrically similar, has similar mass, similar boundary conditions, similar loads, and similar material properties. By applying loads at predetermined locations on the bridge structural model, and by deploying sensors such as laser rangefinders and strain gauges, the deflection and strain at corresponding locations of the bridge structure are measured, the displacement and stress changes of the bridge structure are monitored, the internal forces of the bridge are calculated, and the influence lines of the internal forces of the bridge structure are plotted. The most unfavorable load locations of the bridge are analyzed, and the structural mechanics stress calculations and safety assessments of the bridge are performed.

[0048] S04. Conduct bridge structural load tests, and perform structural mechanical stress calculations and safety assessments on the bridge.

[0049] In one embodiment of the present invention, the stress on the bridge structure is analyzed by conducting bridge load tests on actual long-span bridges in urban areas. By applying loads to specific locations on the bridge structure and deploying sensing devices such as total stations, displacement sensors, full-bridge sections, and rigid-chord strain gauges at specific locations on the bridge deck, towers, and piers, or by deploying peripheral devices (such as mobile phones with built-in acceleration sensors), the mid-span displacement and stress changes at corresponding locations are monitored under load conditions. The most unfavorable load location is analyzed, and the mechanical stress calculation and safety assessment of the long-span bridge structure are performed.

[0050] S05. Establish a seismic response model for long-span bridges, analyze the vibration mechanism, and predict the seismic response.

[0051] In one embodiment of the present invention, a seismic response model for long-span bridges is established using the following three methods: 1. Bridge shaking table test based on bridge scale model / full-scale model of local components - online simulation; 2. Finite element numerical simulation of bridge seismic response based on the finite element method; 3. Bridge forced vibration response test.

[0052] In practice, corresponding experiments should be designed based on the three methods mentioned above: 1. Bridge shaking table experiment based on a scaled-down model of a bridge: First, a bridge structural model with similar geometry, mass, boundary conditions, loads, and material properties is established. By placing the bridge structural model on a bridge shaking table and inputting seismic waves of different types and frequencies, the vibration mode analysis of the bridge structural model under different types and frequencies of seismic waves is carried out.

[0053] 2. Finite element numerical simulation of bridge seismic response based on the finite element method: A finite element analysis model of bridge seismic response is established using appropriate structural analysis software (such as SAP2000), and the stress of the bridge structure under different seismic loads is verified.

[0054] 3. Forced vibration response test of bridges: Excitation devices are installed on actual long-span bridges, and excitation forces are applied to the structure to induce forced vibrations. The dynamic characteristics and vibration mechanism of the bridge structure are analyzed through the resonance response of excitation forces at different frequencies generated by the excitation devices.

[0055] S06. Establish a wind-induced vibration response model for long-span bridges, analyze the principle of wind-induced vibration, and predict the wind-induced vibration response.

[0056] In one embodiment of the present invention, a wind-induced vibration response model for long-span bridges is established using the following three methods: 1. Wind tunnel experiments on bridges based on scaled-down models 2. Finite element numerical simulation of bridge wind vibration response based on computational fluid dynamics (CFD) method 3. Conduct on-site measurements, analyze the principle of wind-induced vibration, and predict the wind-induced vibration response.

[0057] In practice, corresponding experiments should be designed based on the three methods mentioned above: 1. Wind tunnel experiments on bridges based on scaled-down models: First, a bridge structural model with similar geometry, mass, boundary conditions, loads, and material properties is established. The bridge structural model is placed in a wind tunnel, and wind of different types and frequencies is input. Wind pressure and wind speed sensors are installed on the bridge structural model to analyze the wind vibration modes of the bridge model.

[0058] 2. Establish a finite element numerical simulation of bridge wind vibration response based on computational fluid dynamics and structural dynamics analysis, and simulate the bridge response under natural wind force through computer simulation.

[0059] 3. By deploying wind pressure and wind speed sensors on the actual bridge structure surface to collect wind speed and wind pressure data around the bridge, monitor changes in the flow field around the bridge, and establish a bridge wind vibration response model.

[0060] S07. Integrate the stress calculation data, seismic response data, and wind vibration response data of long-span bridge structures obtained by model tests, numerical simulations, and field measurements. Construct a hidden layer between the input and output using deep learning algorithms to establish the inherent logical relationship of the dataset. Based on this, establish a high-fidelity digital twin of long-span bridges for real-time data acquisition and prediction processing.

[0061] S08. Based on a high-fidelity digital twin, an intelligent closed-loop control digital twin system is formed, integrating intelligent monitoring, intelligent data transmission, intelligent data storage, intelligent data analysis, and intelligent structural defect prediction and treatment. This includes a large-span bridge intelligent digital twin system encompassing benchmark modeling, ubiquitous sensing, virtual-real interaction, a high-fidelity digital twin system (accurate prediction), forward design, digital construction, full-state operation and maintenance, and a closed-loop control module. This system aims to achieve automated perception, automated simulation, automated learning, automated evaluation, automated prediction, and automated execution throughout the entire bridge operation process.

[0062] S09. Continuously collect monitoring data and continuously build a digital twin of a long-span bridge that reflects the latest status of bridge construction and operation through the above steps S01 to S08. By comparing the error between the calculated value and the measured value of the digital twin, the digital twin is continuously corrected, thereby realizing the construction and iterative optimization evolution of a digital twin system for intelligent construction and emergency management of urban bridges.

[0063] Establish an intelligent monitoring system based on edge sensors and drone payloads, and a decision-making system for real-time structural status assessment, defect prediction, and operation and maintenance of bridges based on deep learning and artificial intelligence algorithms (AI Agent).

[0064] In one embodiment of the present invention, such as Figure 3 As shown, a digital twin system and method for intelligent construction, operation, maintenance and emergency management of urban bridges are constructed based on the above-mentioned process. The system is used for intelligent construction and operation management of long-span urban bridges. The system is modularly designed and includes eight modules: a baseline modeling module, a ubiquitous sensing module, a virtual-real interaction module, a high-fidelity digital twin system (accurate prediction) module, a forward design module, a digital construction module, a full-state operation and maintenance module and a closed-loop control module. The eight modules cover the entire process of bridge construction and operation management.

[0065] The benchmark modeling module is used to establish corresponding digital models using modern digital modeling technologies such as photogrammetry and laser point cloud to represent the stress characteristics of the bridge structure, the bridge geometry, the bridge hydrological and geological environment, and the bridge's seismic and wind vibration responses under seismic and wind loads.

[0066] The ubiquitous sensing module is used to acquire data such as stress and strain of bridge structural components. By deploying corresponding edge devices, sensors, total stations, etc. on the surface of the bridge structure, a bridge monitoring system is established to monitor the bridge's environmental condition, operational load, and static and dynamic response. Interferometric synthetic aperture radar (InSAR) mounted on platforms such as satellites and UAVs can be used to achieve integrated and precise deformation monitoring of long-span bridges from the air, water, and ground. The measurable data includes the bridge structure's three-dimensional displacement and time-varying characteristics. The ubiquitous sensing module includes an environmental condition monitoring module, a construction and operation load monitoring module, and a bridge structure static and dynamic response monitoring module. The environmental condition monitoring module is used to monitor temperature, wind direction, wind speed, ground motion, bottom scour, seawater corrosion of bridge components and the resulting concrete expansion and cracking. The operation load monitoring module is used to monitor highway vehicle load, lane load, and pedestrian load. The static and dynamic response monitoring module is used to monitor the overall geometric deformation of the bridge structure, the stress distribution and changes of local components, the internal force of the main arch of the arch bridge, the cable force of the stay cables, and the tension of the main cable / suspender of the suspension bridge.

[0067] The virtual-real interaction module is used to realize digital mapping and two-way interaction between the bridge's digital twin and the actual physical bridge, as well as its full-scale (scaled-down) model. It enables real-time assessment of the stress state of the physical bridge through the digital twin, guiding actual construction. It embodies the digital mapping and two-way interaction functions between the digital twin and the physical entity. Rapid information transmission between the digital twin, the actual bridge, and the full-scale (scaled-down) model is achieved through 5G / 5G-A / 6G mobile communication. Through the two-way transmission and interaction of data predicted by the digital twin system with sensor data from the actual bridge and the full-scale (scaled-down) model, the stress state of the bridge can be predicted in real time, ensuring construction safety.

[0068] The high-fidelity digital twin system (accurate prediction) module is used to characterize the geometric, physical, temporal, behavioral, rule-based, evolutionary, evolutionary, update, redundancy, failure, and uncertainty quantification aspects of the benchmark modeling module and the ubiquitous sensing module from multiple dimensions of constitutive relations, multiple time scales, multiple spatial scales, and material-level, component-level, part-level, system-level, and system-level construction, reinforcement, and maintenance post-update methods. The high-fidelity digital twin system (accurate prediction) module includes a reduced-order structural simulation model from the model-driven approach, and a series of data from the data-driven approach, including geometric, physical, temporal, behavioral, loading / unloading history information, dimensionality reduction calculation schemes, and result output definitions. By integrating the model data from the model-driven and data-driven approaches, and using dynamic graph neural networks and graph deep learning methods, the high-fidelity digital twin system (accurate prediction) module constructs representative features of the learning data across multiple processing layers, explores complex patterns and correlations between input and output data, and performs real-time calibration and correction of the digital twin model using intelligent monitoring data, thereby achieving the assessment and prediction of the bridge structure's health status. Meanwhile, the high-fidelity digital twin system (precise prediction) module can use the latest data from the actual operation and maintenance process of long-span bridges collected in real time by devices such as UAV visible light, microwave photogrammetry, laser scanning, and various edge intelligent sensors to calibrate and correct the digital twin model. Through model iteration, the accuracy and prediction accuracy of the digital twin model are continuously improved to achieve the goal of accurate prediction. The high-fidelity digital twin system (precise prediction) module includes a digital twin model calibration and correction module, a digital twin model module, a structural status assessment module, and a bridge defect prediction module. The digital twin model calibration and correction module is used to perform real-time iterative updates of the digital twin based on the data monitored by the ubiquitous sensing module. The digital twin model module includes a geometric model integrating BIM and GIS, a point cloud model generated by laser scanning and photogrammetry, and a fusion information model obtained by microwave measurement, as well as a finite element model including bridge poles and solid elements, a CFD simulation model reflecting the hydrological environment (HIM) and airflow field (AIM) around the bridge, and a geotechnical engineering seismic response simulation model reflecting the seismic wave field information model (EIM). The structural status assessment module is used to perform real-time analysis and assessment of the internal forces and deformations generated by external loads in the bridge structure. The bridge defect prediction module is used to predict potential defects in the bridge structure (such as concrete cracking, excessive structural deformation, etc.) in real time. The technologies used include cloud computing, blockchain, big data, Internet of Things, photogrammetry, laser scanning, large AI models, model reduction, virtual simulation, machine learning, computer vision, deep learning, artificial intelligence (AI), edge devices, augmented reality (AR), and virtual reality (VR).

[0069] The forward design module is used to monitor the stress on the bridge during construction using a pre-built digital twin, and updates the digital twin in stages as construction progresses. This dynamically changing digital twin can guide on-site bridge construction through real-time predictions and monitor the bridge's stress characteristics in real time, ensuring safety during construction. The forward design module can also address quality control issues during the actual construction of long-span bridges. This module can convert the point cloud model of the steel / concrete box girder into a three-dimensional solid model through the benchmark modeling module in the digital twin system, and simulate the pre-assembly of the steel / concrete box girder in the virtual environment of the digital twin system. This enables intelligent construction of long-span urban bridges through virtual-real interaction and construction between the digital twin and the physical entity.

[0070] The digital construction module is used to control digital construction machinery for automated operation after the forward design module is completed. In one embodiment of the present invention, it is used to control unmanned hanging baskets, unmanned cranes, intelligent prestressed tensioning equipment, intelligent winches, steel mesh robots, concrete pouring robots and concrete curing robots for automated construction through the digital twin system. (1) The digital twin system is used to automatically cast concrete beam segments and automatically hoist steel box girders using intelligent unmanned hanging baskets and unmanned cranes; (2) The intelligent prestressed tensioning equipment is used to automatically tension prestressed steel bars and anchor the tensioning ends; (3) The intelligent jack equipment is used to automatically tension the stay cables; (4) During the secondary paving of the bridge deck, the digital twin system is used to command the steel mesh robots, concrete pouring robots and concrete curing robots to perform a series of automated construction steps such as steel mesh binding, concrete pouring and concrete watering curing.

[0071] The comprehensive operation and maintenance module integrates multi-dimensional information on the entire modality and state of bridge construction and operation, enabling intelligent management and control throughout the entire lifecycle of bridge construction and operation. The bridge digital twin system encompasses multi-dimensional information on the entire modality and state of bridge construction and operation. The comprehensive modality includes information carriers such as sound, light, electricity, text, images, and video related to the bridge structure's construction, operation, maintenance, and emergency management processes. The state information represents the spatiotemporal stage of the bridge structure. Through the integration of comprehensive bridge information, the digital twin system can perform predictive maintenance and intelligent management and control throughout the entire lifecycle of bridge construction and operation.

[0072] The closed-loop control module provides real-time risk management strategies for the operation of long-span bridges. Monitoring data collected by edge devices and sensors deployed on the bridge surface is uploaded via the internet to a high-fidelity digital twin system (precise prediction) module on the server. This module assesses and predicts the health status of the bridge structure. The closed-loop control module also features safety risk control, emergency management, and automatic execution functions. For example, it can automatically issue warnings or urgently evacuate people and vehicles when excessive wind speeds are detected in the flow field around the bridge by the ubiquitous sensing module and when the bridge's wind-induced vibration exceeds the standard, based on predictions from the precise prediction module. It can also automatically restrict traffic flow using traffic lights when the bridge deck is overloaded, and notify maintenance personnel to repair and reinforce the bridge when cracks are detected by drones. The closed-loop control module uses an artificial intelligence agent and corresponding actuators deployed on the actual bridge surface to intelligently make decisions and regulate the behavior and operating status of personnel, machines, materials, methods, and the environment during the construction, operation, and maintenance of long-span bridges, achieving real-time warning, real-time feedback, and real-time control. The AI ​​Agent is trained using deep learning (such as dynamic graph neural networks and graph deep learning methods). The AI ​​Agent proposes suggestions and countermeasures for bridge risk management, and carries out safety risk control and emergency management. Practical applications include: (1) real-time warning when potential risks to the bridge are detected, including but not limited to changing the construction technology organization and process flow, reinforcing construction equipment, and emergency evacuation of construction personnel and vehicles; (2) real-time feedback to bridge traffic signal control and warning equipment, GNSS navigation equipment for passing vehicles, public instant messaging platforms, etc., and controlling the traffic flow of vehicles on the bridge / emergency closure of bridge traffic; (3) real-time feedback to the bridge operation and maintenance department so that bridge engineers can organize personnel to carry out bridge reinforcement and maintenance work in a timely manner.

[0073] The high-fidelity digital twin system (accurate prediction) module also includes an AI agent based on a large AI model. The results of the aforementioned assessment and prediction can be automatically analyzed by the AI ​​agent, which will then automatically propose corresponding suggestions for risk management of long-span bridges, thereby achieving intelligent decision-making. The AI ​​agent is an application software package driven by a pre-trained large AI model. The pre-training set includes predicted values ​​of stress characteristics obtained from material parameter calculations in forward analysis, test values ​​from scaled-down / full-scale bridge structure models, and material parameters derived from on-site monitoring stress characteristic values ​​in inverse analysis. By optimizing the natural environment, structural geometry and mechanical parameters, and static and dynamic performance parameters of urban bridge construction, operation, maintenance, and emergency management processes, and through AI pre-training, sub-products of different modalities (text, images, videos) can be derived during the iteration process, namely, large AI models (hundreds of billions of neurons), medium AI models (hundreds to tens of billions of neurons), and small AI models (hundreds of thousands to hundreds of neurons). Based on this, various AI models can automatically analyze the results of the above assessment and prediction in combination with different specific application scenarios (specific parameters) and provide relevant industry personnel with corresponding suggestions for risk management of long-span bridges, thereby achieving the purpose of intelligent decision-making.

[0074] The benchmark modeling module employs modern digital modeling technologies such as spaceborne and airborne photogrammetry, and laser point cloud analysis to establish corresponding digital models. These include finite element models (FEM) to represent the structural stress characteristics of bridges; building information models (BIM) to represent the geometric structure of bridges; geological information models (GIS) to represent the hydrological and geological environment of bridges; and bridge dynamic models (FEM+CFD), hydrodynamic information models (HIM), airflow field information models (AIM), and seismic wave field information models (EIM) to represent the seismic and wind-induced vibration responses of bridges under seismic and wind loads. When establishing a precise bridge point cloud model, a UAV equipped with optical cameras, thermal imaging cameras, lidar, and microwave radar can automatically plan the UAV flight path using the pre-established bridge BIM, achieving precise UAV modeling of the bridge. The resulting high-precision point cloud model (millimeter-level) can be used to establish a refined three-dimensional solid model of the bridge through artificial intelligence, 3D reconstruction, and plane fitting technologies.

[0075] The closed-loop control module also features safety risk control, emergency management, and automatic execution functions. For example, it can detect excessive wind speeds in the flow field around the bridge using a ubiquitous sensing module, and automatically issue warnings or urgently evacuate people and vehicles when the bridge's wind-induced vibration exceeds the standard, based on a precise prediction module. It can also automatically restrict traffic flow using traffic lights when the bridge deck is overloaded, and notify maintenance personnel to carry out repairs and reinforcement when a drone detects cracks in the bridge deck. This closed-loop control module needs to intelligently make decisions and regulate the behavior and operating status of personnel, machines, materials, methods, and the environment during the construction, operation, and maintenance of long-span bridges, achieving real-time early warning, real-time feedback, and real-time control.

[0076] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A digital twin system for intelligent construction, operation, maintenance, and emergency management of urban bridges, characterized in that, It includes a benchmark modeling module, a ubiquitous sensing module, a virtual-real interaction module, a high-fidelity digital twin system module, a forward design module, a digital construction module, a full-state operation and maintenance module, and a closed-loop control module; The benchmark modeling module is used to build digital models; The ubiquitous sensing module is used to acquire stress and strain data of bridge structural components and to monitor bridge environmental conditions, operational loads, and static and dynamic responses. The virtual-real interaction module is used to realize the digital mapping and two-way interaction between the digital twin and the physical entity; The high-fidelity digital twin system module is used to characterize and describe the geometric, physical, temporal, behavioral, rule, evolution, evolution, update, redundancy, failure, and uncertainty quantification of the benchmark modeling module and ubiquitous sensing module from multiple dimensions of constitutive relations, multiple time scales, multiple spatial scales, and material level, component level, part level, system level, and system level construction, reinforcement and maintenance post-system update methods. It constructs representative features that conform to the data-driven processing layer of the structural and stress characteristics of long-span bridges through dynamic graph neural networks and graph deep learning methods, builds complex patterns and correlations between input and output data, and performs real-time calibration and correction of the digital twin model through intelligent monitoring data. The forward design module is used to monitor the stress on the bridge through the ubiquitous sensing module during bridge construction using a pre-established digital twin, and to update the digital twin in stages as construction progresses. The digital construction module is used to control the automated operation of digital construction machinery through a digital twin system. The full-state operation and maintenance module is used to integrate multi-dimensional information on the full mode and full state of bridge construction and operation and maintenance, and to perform predictive maintenance and intelligent management and control of the entire life cycle of bridge construction and operation and maintenance. The closed-loop control module is used to provide real-time risk management and control measures for the construction and operation phase of long-span bridges. The monitoring data is uploaded to the high-fidelity digital twin system module on the server side, and the high-fidelity digital twin system module is used to assess and predict the health status of the bridge structure.

2. The digital twin system for intelligent construction, operation, maintenance, and emergency management of urban bridges according to claim 1, characterized in that, The digital model includes a full-state simulation model of the physical system, an AI-based reduced-order model of the physical system, and a geometric model.

3. The digital twin system for intelligent construction, operation, maintenance, and emergency management of urban bridges according to claim 1, characterized in that, The ubiquitous sensing module includes an environmental condition monitoring module, a construction and operation load monitoring module, and a bridge structure static and dynamic response monitoring module. The environmental condition monitoring module is used to track and monitor factors that directly affect the corrosion of steel and reinforcing bars in bridge components and the resulting defects such as concrete expansion, cracking, damage, spalling, and spalling, as well as the annual average temperature and solar temperature gradient, wind speed, wind pressure, earthquakes, frost, rain, and snow. The operational load monitoring module is used to monitor highway vehicle load, lane load, and pedestrian load; The static and dynamic response monitoring module is an edge-cloud integrated equipment used to monitor the overall structural geometric deformation of bridges, stress distribution and changes in local components, internal forces of the main arch of arch bridges, cable force of stay cables, tension of main cables / suspenders of suspension bridges, vibration isolation bearings, and dampers. It also meets the requirements of intelligent data processing, data-driven operation, real-time management and disaster backup for edge monitoring equipment.

4. The digital twin system for intelligent construction, operation, maintenance, and emergency management of urban bridges according to claim 1, characterized in that, The high-fidelity digital twin system module collects the latest data from the actual construction and operation of long-span bridges in real time to calibrate and correct the digital twin model, and continuously improves the accuracy and prediction accuracy of the digital twin model through model iteration.

5. A digital twin system for intelligent construction, operation, maintenance, and emergency management of urban bridges according to claim 1, characterized in that, The high-fidelity digital twin system module includes a digital twin model calibration and correction module, a digital twin model module, a structural status assessment module, and a bridge defect prediction module. The digital twin model calibration and correction module is used to perform real-time iterative updates of the digital twin based on the data monitored by the ubiquitous sensing module; The structural condition assessment module is used to perform real-time analysis and assessment of the internal forces and deformations generated by external loads in bridge structures. The bridge defect prediction module is used to predict potential defects in bridge structures in real time.

6. A digital twin system for intelligent construction, operation, maintenance, and emergency management of urban bridges according to claim 5, characterized in that, The high-fidelity digital twin system module also includes an artificial intelligence agent. The assessment and prediction results are intelligently analyzed by the artificial intelligence agent, which automatically proposes corresponding suggestions for risk management of long-span bridges.

7. A digital twin system for intelligent construction, operation, maintenance, and emergency management of urban bridges according to claim 1, characterized in that, The forward design module is also used for monitoring the stress state of the bridge and iteratively updating the model.

8. A digital twin system for intelligent construction, operation, maintenance, and emergency management of urban bridges according to claim 1, characterized in that, The ubiquitous sensing module is also used for integrated air, space, and ground deformation monitoring of long-span bridges, and the measured data includes the deflection and displacement at the center of the bridge span.

9. A digital twin system for intelligent construction, operation, maintenance, and emergency management of urban bridges according to any one of claims 1-8, characterized in that, The closed-loop control module uses an artificial intelligence agent trained by deep learning to automatically generate intelligent operation and maintenance control suggestions and countermeasures for bridge risk management. After manual confirmation, safety risk control and emergency management are carried out.

10. A digital twin method for intelligent construction and emergency management of urban bridges, characterized in that, Includes the following steps: Establish a 3D point cloud building information model and geological information model based on UAV lidar and photogrammetry technology; Establish a finite element calculation model based on refined BIM and GIS, perform structural mechanical stress verification and safety assessment of the bridge, and regularly revise the finite element model using collected monitoring data; Conduct loading tests on bridges to perform structural mechanics stress calculations and safety assessments on the bridges; Conduct load tests on bridge structures, and perform structural mechanical stress calculations and safety assessments on the bridges. Establish a seismic response model for long-span bridges, analyze the vibration mechanism, and predict the seismic response; Establish a wind-induced vibration response model for long-span bridges, analyze the principle of wind-induced vibration, and predict the wind-induced vibration response; By integrating structural stress calculation data, wind vibration response data, and seismic response data of long-span bridges, and using data-driven deep learning algorithms, the inherent logical relationship of the dataset is established. Monitoring data on the operational status of long-span bridges is collected in real time to predict and assess the safety status of bridges in real time. Based on this, a high-fidelity digital twin of long-span bridges is established for real-time data collection and prediction processing. Based on a high-fidelity digital twin, an intelligent closed-loop control digital twin system is formed, integrating intelligent monitoring, intelligent data transmission, intelligent data storage, intelligent data analysis, and intelligent structural disease prediction and treatment. By continuously collecting monitoring data and comparing the errors between the calculated and measured values ​​of the digital twin, the digital twin is constantly corrected, thereby realizing the construction and iterative optimization of a digital twin system for intelligent construction and emergency management of urban bridges.

Images