Railway bridge hierarchical early warning method and system

Abstract

The present application relates to the technical field of railway bridges, in particular to a hierarchical early warning method and system for railway bridges, comprising a digital twin model mapping the whole life cycle state of a physical bridge; multi-source environmental data such as geographical geology, hydrology, meteorology, etc. are obtained to generate multi-level load working conditions; the model is loaded for multi-physical field coupling simulation to obtain multi-scale structural response data of the bridge as a whole, the superstructure and substructure, and key components; after performance state evaluation, an early warning level and optimization strategy are generated according to a hierarchical early warning rule base, and a report containing the early warning level, risk location, causes and optimization strategy is output. The present application can predict future complex environments and load actions in the design stage by creating a digital twin, be aware of risks in advance, change the traditional post-alarm mode, form an intelligent closed loop of "simulation-early warning-optimization-verification", and ensure the long-term safety of railway bridges and reduce the whole life cycle maintenance cost.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of railway bridges, in particular to a hierarchical early warning method and system for railway bridges. BACKGROUND

[0002] As the throat of the railway network, the safety and durability of railway bridges are of great importance. The traditional safety management of railway bridges mainly relies on two stages: theoretical calculation in the design stage and health monitoring in the construction and operation stage.

[0003] In the design stage, engineers mainly use finite element analysis and other calculation methods to perform static and dynamic analysis on the bridge structure according to the specifications; however, this analysis method has significant limitations:

[0004] Firstly, the calculation model is usually a simplified ideal model, which is difficult to truly reflect the complex coupled system composed of the superstructure, bearings, substructure and surrounding geological and hydrological environment of the bridge; secondly, the load conditions are mostly standardized isolated conditions, which cannot simulate the combined action and cumulative effect of various environmental loads (such as wind, rain, temperature, water erosion, foundation settlement, etc.) that the bridge actually bears over its entire life cycle and dynamically changes over time. This leads to some potential risks, especially performance degradation under extreme weather or long-term environmental erosion, which is difficult to discover in the design stage.

[0005] Moreover, although bridge health monitoring systems have been widely used in the operation stage, they monitor the stress, displacement, vibration, etc. of the bridge in real time through the arrangement of sensors and achieve over-limit alarm. However, this kind of system is essentially a passive response mechanism after the event or during the event. Only when the structure has already appeared damage or abnormal response, the system will alarm, at this time the structure may have entered an unsafe state, the cost of repair and reinforcement is high, and the huge economic loss caused by operation interruption cannot be avoided.

[0006] In view of this, a hierarchical early warning method and system for railway bridges are proposed. SUMMARY

[0007] The purpose of the present application is to provide a hierarchical early warning method and system for railway bridges to solve the problem of limited model analysis capability in the design stage of railway bridges, which is difficult to support complex performance prediction and safety warning.

[0008] To solve the above technical problems, the present application provides a hierarchical early warning method for railway bridges, comprising the following steps:

[0009] S1, constructing a digital twin model of the railway bridge; the digital twin model is a virtual model that maps the full life cycle state of the physical bridge;

[0010] S2. Acquire multi-source environmental data of the target bridge site area, and generate multi-level load conditions based on the multi-source environmental data; the multi-source environmental data includes geographic geological data, hydrological data, and meteorological data;

[0011] S3. Load the multi-level load conditions onto the digital twin model and perform multi-physics coupling simulation calculations to obtain the structural response data of the bridge at multiple scales; the multiple scales include the overall bridge, superstructure, substructure and key support components.

[0012] S4. Based on the structural response data, evaluate the performance status of the bridge, and generate corresponding early warning levels and optimization strategies according to the preset hierarchical early warning rule base to form an early warning report; the hierarchical early warning rule base includes multiple levels from suggestive early warning to catastrophic early warning.

[0013] S5. Output an early warning report, which includes the early warning level, risk location, risk cause, and recommended optimization strategy.

[0014] As a further improvement to this technical solution, the step S1 of constructing a digital twin model of the railway bridge specifically includes:

[0015] Based on the bridge design drawings, a geometric model of the bridge is established;

[0016] The geometric model is assigned material properties, boundary conditions, and connection relationships to form a structural computational model containing multi-physics properties.

[0017] Integrate the basic geographic information model of the target bridge site area to complete the construction of a digital twin model.

[0018] As a further improvement to this technical solution, step S2 generates multi-level load cases, specifically including:

[0019] Based on historical data and probability models, a continuous load spectrum is generated from normal use to extreme adverse events.

[0020] The load spectrum includes load conditions for assessing river scour and foundation settlement of the substructure, load conditions for assessing wind and temperature loads of the superstructure, and load conditions for assessing train dynamic loads of the overall structure.

[0021] As a further improvement to this technical solution, step S3 involves multiphysics coupling simulation calculations, specifically including:

[0022] Perform fluid-structure interaction simulation to simulate the interaction between water flow and bridge pier structure;

[0023] Perform soil-structure interaction simulation to model the interaction between the foundation and the bridge foundation;

[0024] Calculate the stress, strain, displacement, and vibration response of the bridge at each scale level under the aforementioned multi-level load conditions.

[0025] As a further improvement to this technical solution, step S4 generates warning levels and optimization strategies based on a preset hierarchical warning rule base, specifically including:

[0026] The structural response data obtained from the simulation is compared with a preset safety threshold;

[0027] If a key indicator exceeds the safety threshold, the corresponding warning level will be triggered based on the scope and degree of the exceedance.

[0028] The optimization strategy is generated from a pre-set strategy library based on the warning level, and the strategy content includes design parameter adjustment, structural form change and bridge site reselection.

[0029] As a further improvement to this technical solution, the warning levels include:

[0030] A Level 1 warning indicates that the indicator is approaching its threshold and should be monitored.

[0031] A Level 2 general warning indicates that the response of a local component exceeds the limit, and design optimization is recommended.

[0032] A Level 3 severity warning indicates that there is a risk of damage to critical components and recommends major design changes.

[0033] A Level 4 catastrophic warning indicates that the structure is at risk of instability and collapse, and recommends that the current plan be rejected.

[0034] As a further improvement to this technical solution, the method further includes step S6:

[0035] Based on the optimization strategy output in step S5, the bridge design is modified, and steps S1 to S5 are repeated for iterative simulation verification. When the simulation results no longer trigger the Level 3 severity warning, and all Level 2 general warnings are eliminated or their number is lower than the preset threshold, the design optimization loop is completed.

[0036] A graded early warning system for railway bridges, the graded early warning system for railway bridges being used to implement the aforementioned graded early warning method for railway bridges, comprising:

[0037] The data acquisition and integration module is used to receive bridge design data and multi-source environmental data;

[0038] The digital twin model building module, connected to the data acquisition and integration module, is used to build and update digital twin models of railway bridges.

[0039] The simulation calculation module, connected to the digital twin model construction module, is used to perform multiphysics coupling simulation calculations.

[0040] The early warning analysis and decision-making module is connected to the simulation calculation module and contains the hierarchical early warning rule library, which is used to analyze simulation results and generate early warning reports.

[0041] The visualization and interaction module, connected to the early warning analysis and decision-making module, is used to display the simulation process, early warning information, and optimization strategies.

[0042] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0043] 1. The graded early warning method and system for railway bridges creates a "digital twin" covering the entire lifecycle of railway bridge design, construction, and operation. Through multi-physics coupling simulation, it can predict various complex environments and loads that the bridge may experience in the coming decades at the design blueprint stage, achieving "predictive awareness" of potential risks. At the same time, it moves the safety management checkpoint forward, fundamentally changing the traditional post-event alarm mode and forming an intelligent closed loop of "simulation-early warning-optimization-verification". This significantly improves the scientific nature and reliability of bridge design, ensures the long-term safety of railway bridges from the source, and effectively reduces the maintenance cost throughout the entire lifecycle.

[0044] 2. The graded early warning method and system for railway bridges integrates multi-source environmental data such as geography, geology, hydrology, and meteorology, and constructs a complete digital twin model including the superstructure, substructure, and supports. This enables full-system, multi-scale simulation of the bridge and its surrounding environment, overcoming the limitations of traditional simplified models and making early warning judgments more comprehensive and accurate.

[0045] 3. The graded early warning method and system for railway bridges can pre-simulate multi-level load conditions ranging from normal use to extreme conditions in virtual space, thereby identifying hidden safety risks that only appear under specific combinations of adverse conditions. This allows for targeted reinforcement during the design phase, preventing these risks from erupting during the actual bridge operation phase.

[0046] 4. The graded early warning method and system for railway bridges establishes multi-level early warning rules ranging from suggestive to catastrophic warnings, and intelligently associates different levels of early warnings with specific design optimization strategies (such as parameter adjustment, structural changes, and bridge site reselection), providing designers with clear and quantitative decision-making basis and significantly improving the efficiency and scientific nature of design optimization. Attached Figure Description

[0047] Figure 1 This is a flowchart illustrating the overall method of the present invention;

[0048] Figure 2This is an overall system block diagram of the present invention. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] Example 1

[0051] Currently, the shortcomings of traditional bridge design are as follows:

[0052] 1. The digital model is only a simplified geometric model and does not incorporate material properties, boundary conditions and geographical environment, so it cannot simulate the complex coupled system of "bridge-environment";

[0053] 2. The load conditions are isolated standardized values ​​(such as considering only static train loads), which cannot reflect the cumulative effects of dynamic loads such as wind, water, and temperature throughout the entire life cycle;

[0054] 3. Without multiphysics coupling simulation, it is impossible to calculate the interaction between water flow and bridge piers, and between foundation and substructure;

[0055] 4. The early warning system is a "post-event alarm," lacking tiered early warning and a closed-loop design optimization mechanism, making it difficult to avoid potential risks in advance;

[0056] In view of this, please refer to Figure 1 As shown, one of the objectives of this embodiment is to provide a graded early warning method for railway bridges. This method, using a 32m simply supported T-beam railway bridge (design speed 160km / h, design service life 100 years) as an example, will be explained in detail, and specifically includes the following steps:

[0057] Because traditional finite element models only include geometry, material properties are taken as standard averages, and boundary conditions are set as the "fixed end" of graded early warning for railway bridges, without integrating bridge site geographic information, they cannot reflect the structural response under actual working conditions. Therefore:

[0058] S1. Constructing a digital twin model of a railway bridge; the digital twin model refers to a virtual model that maps to the physical bridge's full life-cycle state; specifically, constructing a digital twin model of a railway bridge includes: establishing a geometric model of the bridge based on bridge design drawings; assigning material properties, boundary conditions, and connection relationships to the geometric model to form a structural calculation model containing multi-physics field properties; integrating the basic geographic information model of the target bridge site area to complete the construction of the digital twin model;

[0059] The specific steps are as follows:

[0060] First, obtain the complete set of design drawings for the 32m simply supported T-beam bridge (including elevation drawings, section drawings, reinforcement layout drawings, and foundation structure drawings, in Auto CADDWG 2020 format), and then use Autodesk Revit 2023 software to model it at a 1:1 scale; thus establishing the bridge's geometric model.

[0061] Specific parameters include: Superstructure: T-beam web thickness 200mm, flange width 1600mm, beam height 2000mm, with 5 T-beams per span and a beam spacing of 300mm; Substructure: Circular piers with a diameter of 1.2m and a height of 8m, with enlarged foundation dimensions of 4m×4m×1.5m; Bearings: GYZ300×63 type rubber bearings, with upper and lower connecting plates made of Q235 steel, 20mm thick. Simultaneously, the coordinate system was calibrated during modeling to match the GNSS coordinates at the bridge site (e.g., 118°25′E, 30°12′N) to ensure spatial location matching.

[0062] Secondly, the geometric model is imported into ANSYS AIM 2023 software, and material properties are assigned according to component type (data source: material factory inspection reports and on-site sampling tests), thereby assigning multiphysics properties and boundary conditions. The specific content includes the following:

[0063] For the construction of T-beams and piers, the assigned material type is C50 concrete with a density of 2400 kg / m³. 3 The elastic modulus is 3.45 × 10⁻⁶. 4 The axial compressive strength is 23.1 MPa, with a Poisson's ratio of 0.2. For the reinforcing steel, the assigned material type is HRB400, with a density of 7850 kg / m³. 3 The elastic modulus is 2.0 × 10⁻⁶. 5 MPa, Poisson's ratio is 0.3, and yield strength is 402 MPa; for the construction of rubber bearings, the material is natural rubber with a density of 1100 kg / m³. 3 The elastic modulus is 1.5 × 10⁻⁶. 3 MPa, Poisson's ratio is 0.45, and shear modulus is 0.5 × 10⁻⁶. 3 MPa; For the construction of foundations and anti-slide piles, the assigned material type is C30 concrete with a density of 2350 kg / m³. 3 The elastic modulus is 3.0 × 10⁻⁶. 4 MPa, Poisson's ratio is 0.2, and the foundation bearing capacity is 300 kPa;

[0064] Then, boundary conditions were set (referencing the "Code for Design of Railway Bridge and Culvert Foundations" TB 10093-2017), where the bottom of the pier and the foundation adopted "bound constraint"; the foundation and the ground adopted "elastic support", and the spring stiffness was determined through field load tests (using the slow sustained load method, with a bearing plate area of ​​0.5m²). 2 The measured proportional limit load of the pebble layer is 250 kPa. According to the formula k = (π × d × Es) / (4 × (1 - μ)), 2 Calculations show that d = 0.798m (diameter of the bearing plate), Es = 18MPa (compression modulus), and μ = 0.3 (Poisson's ratio), resulting in a final spring stiffness of 1.2 × 10⁻⁶. 5 N / m, the formula is defined as follows: k is the spring stiffness (N / m), d is the spring wire diameter (mm), Es is the elastic modulus of the spring material (MPa), and μ is Poisson's ratio); the support and the beam and pier adopt "binding constraint" to simulate the actual connection state;

[0065] Finally, GIS data at a scale of 1:500 for the bridge site area was obtained (source: engineering geological survey report from the local natural resources bureau), and a basic geographic information model was integrated, including the following:

[0066] Topographic elevation data (DEM resolution 1m, ground elevation at bridge site 20m); stratigraphic distribution (top layer 0-3m is silty clay, 3-8m is pebble layer, below 8m is moderately weathered sandstone); hydrological baseline (normal river level at bridge site 2.5m, riverbed slope 0.5%).

[0067] ArcGIS Pro 3.0 software was used to spatially overlay the above data with the bridge model to ensure that the bridge pier foundation was embedded 2m into the pebble layer, which perfectly matched the actual geological conditions.

[0068] Through the above steps, the constructed digital twin model not only includes geometric shape, but also integrates the real properties of materials, physical boundary conditions and geographical environment, which solves the defects of "simplification and idealization" in traditional models, provides a "high-fidelity" foundation for subsequent multiphysics simulation, and reduces the calculation error of structural response.

[0069] Considering that traditional design only uses isolated loads specified in the standards (such as train static load and basic wind pressure), it cannot simulate the dynamic changes of "normal-extreme" loads and the coupling effects of multiple loads throughout the entire life cycle, leading to the omission of risks under extreme conditions, therefore:

[0070] S2. Acquire multi-source environmental data of the target bridge site area and generate multi-level load cases based on the multi-source environmental data; the multi-source environmental data includes geographic geological data, hydrological data, and meteorological data; wherein, generating multi-level load cases in step S2 specifically includes: generating a continuous load spectrum from normal use to extreme adverse events based on historical data and probability models; the load spectrum includes river scour and foundation settlement load cases for evaluating the substructure, wind load and temperature load cases for evaluating the superstructure, and train dynamic load cases for evaluating the overall structure;

[0071] The specific steps are as follows:

[0072] First, multi-source environmental data was acquired, specifically including: geographic geological data, obtained from the local natural resources bureau by retrieving engineering geological survey reports (e.g., No. GH2024-058), with key parameters (for the bridge site area) including stratigraphic distribution and foundation bearing capacity (e.g., 300 kPa for the pebble layer), updated only once (when there are no changes); and hydrological data, obtained from on-site monitoring and watershed hydrological stations, through the deployment of ultrasonic water level gauges (e.g., RS485-UTG) and current meters (e.g., LS25-1) 500m upstream of the bridge site, synchronously accessing data from the watershed hydrological stations. Key parameters (for the bridge site area) were also acquired. The data for the bridge site area includes monthly average water level, maximum flow velocity, and 100-year return period scour depth, updated every 10 minutes. Meteorological data is obtained from the China Meteorological Administration Public Service Center through API calls (e.g., interface number: CMA-PUB-2024-03), with key parameters (for the bridge site area) including extreme wind speed, extreme temperature, and annual rainfall, updated every 5 minutes. Train load data is obtained from railway operating departments by retrieving train operation records from the past 10 years, with key parameters (for the bridge site area) including daily axle load, heavy-load axle load, and maximum speed, updated monthly.

[0073] Then, using "extreme value theory (Gumbel distribution) and load coupling analysis", a continuous load spectrum of 5 levels is generated, covering the entire scenario from "normal use to extreme adverse conditions", generating multi-level load cases with the following specific parameters:

[0074] For load levels of normal use, the applicable scenario is daily operation (90% of the time). The specific load parameters are: train axle load 25t (speed 120km / h), wind speed 10m / s, water level 2.5m, and temperature 15℃ (annual average). The structural part is the integral structure.

[0075] For the load level of light load, the applicable scenario is monthly extreme load (1-2 times per month). The specific load parameters are train axle load of 27t (speed 160km / h), wind speed of 18m / s, water level of 3.8m, and temperature of 35℃. The structural parts are the superstructure (T-beam) and supports.

[0076] For the load level of medium load, the applicable scenario is the annual extreme load (once a year). The specific load parameters are: train axle load 30t (speed 160km / h), wind speed 22m / s, water level 4.5m, temperature 42℃, and scour depth 1.5m. The applicable structural parts are the superstructure and the substructure (bridge pier).

[0077] For the heavy load level, the applicable scenario is once in 50 years (design reference period). The specific load parameters are: train axle load 32t (speed 200km / h), wind speed 28m / s, water level 5.2m, temperature -10℃, scour depth 2.3m, and the structural parts are the substructure and foundation.

[0078] For the extreme load level, the applicable scenario is a 100-year return period (safety reserve). The specific load parameters are: train axle load 35t (speed 200km / h), wind speed 35m / s, water level 5.5m, temperature -15℃, and scour depth 3.1m. The structural part is the integral structure (anti-instability verification).

[0079] At the same time, specific working conditions are set for different structural parts:

[0080] Substructure: Loading river scour (scour depth gradually changes from 0 to 3.1m to simulate the annual scour effect) and foundation settlement (pebble layer settlement 0 to 50mm to simulate uneven settlement).

[0081] Superstructure: Loaded wind load (longitudinal and transverse directions, wind pressure calculated according to GB 50009-2012, transverse wind pressure 1.2kPa), temperature load (annual temperature difference -15℃ to 42℃, gradient temperature difference 5℃ / m, simulating beam temperature stress).

[0082] Overall structure: Loaded with train dynamic load (using a moving load model, considering vehicle-bridge coupled vibration, wheel-rail contact force calculated according to Hertz contact theory, impact coefficient 1.3);

[0083] By breaking through the limitations of traditional "isolated loads" through the above steps, the generated continuous load spectrum can simulate the dynamic changes of loads throughout the entire life cycle and the coupling effect of multiple loads (such as the combination of "strong wind + heavy-load train + high water level"), and discover hidden risks that are missed in traditional design in advance (such as the excessive tensile stress at the mid-span of T-beams under extreme loads), thereby improving the load coverage.

[0084] Considering that traditional design only performs single-physics field calculations (such as structural statics), it cannot simulate the coupling effect of "water flow-pier" and "foundation-base", resulting in large deviations in the calculation of pier hydrodynamic pressure and foundation reaction force, and failing to reflect the true stress state of the structure, therefore:

[0085] S3. Apply the multi-level load conditions to the digital twin model and perform multi-physics coupled simulation calculations to obtain structural response data of the bridge at multiple scales. These multiple scales include the overall bridge structure, superstructure, substructure, and key bearing components. Specifically, step S3 involves multi-physics coupled simulation calculations, including: performing fluid-structure interaction simulation to simulate the interaction between water flow and the bridge pier structure; performing soil-structure interaction simulation to simulate the interaction between the foundation and the bridge foundation; and calculating the stress, strain, displacement, and vibration response of the bridge at each scale under the multi-level load conditions. The specific operation process is as follows:

[0086] ANSYS Workbench 2023 R2 was used as the simulation platform, integrating the CFX (fluid-structure interaction), FLAC3D (soil-structure interaction), and Mechanical (structural response) modules for co-simulation. The steps are as follows:

[0087] First, perform fluid-structure interaction simulation (simulating the interaction between water flow and bridge piers):

[0088] Centered on pier No. 3, construct a 10m×10m×8m cuboid fluid domain (medium: water, density 1000kg / m³). 3 Dynamic viscosity 1.002×10 -3 The model is established by using a tetrahedral mesh (20mm element size) for the pier surface and a hexahedral mesh (50mm element size) for the fluid domain, resulting in approximately 1.2 million elements. The fluid inlet is set as a "velocity boundary" (according to an extreme load level flow velocity of 3.5m / s, with a parabolic velocity distribution), and the outlet is set as a "pressure boundary" (atmospheric pressure 101.3kPa). The pier surface is set as a "no-slip wall," thus completing the boundary condition settings. The time step is 0.01s, and the total calculation time is 10s. The calculation settings for the dynamic water pressure on the pier surface are recorded every 0.1s. Under extreme load, the maximum dynamic water pressure on the upstream side of the pier is 12kPa, and on the lateral side it is 5kPa. Compared with the traditional calculation based solely on hydrostatic pressure (6kPa), this result is more consistent with the actual stress (dynamic water pressure is 100% higher than hydrostatic pressure).

[0089] Secondly, soil-structure interaction simulation (simulating the interaction between the foundation and the subgrade) is performed:

[0090] A 20m×20m×15m three-dimensional foundation model was constructed using FLAC3D 7.0. Parameters were assigned according to the geological strata distribution (silty clay: cohesion 25kPa, internal friction angle 18°; gravel layer: cohesion 0, internal friction angle 35°; moderately weathered sandstone: cohesion 500kPa, internal friction angle 45°), thus establishing the foundation model. The bridge foundation model was then bound to the foundation model using ANSYS and FLAC3D (using MPC multi-point constraints) to ensure force transmission, thereby completing the coupling setup. Simulations were then performed under extreme load levels. The foundation settlement (50mm settlement on the left side and 30mm settlement on the right side of the pebble layer, simulating the differential settlement caused by uneven scouring) was used to complete the loading condition; the final output results showed that the maximum compressive stress at the bottom of the pier was 15MPa (less than the compressive strength of C50 concrete 23.1MPa) and the maximum tensile stress at the edge of the foundation was 1.2MPa (less than the tensile strength of C30 concrete 1.43MPa). However, the traditional calculation of pier stress without soil-structure coupling had a large deviation (the traditional calculation was 11MPa, while the actual value was 15MPa).

[0091] Finally, the structural response data is calculated:

[0092] Twenty monitoring points were set up at key parts of the bridge (covering the entire structure, superstructure, substructure, and supports). The stress, strain, displacement, and vibration response of each monitoring point were calculated under five load levels. Some key results are as follows (extreme load levels):

[0093] The monitoring point was located at the bottom of the mid-span of the T-beam in span 2. The monitoring index was the maximum tensile stress, with a measured value of 3.2 MPa and a traditional calculated value of 2.1 MPa, resulting in a deviation rate of +52%.

[0094] The monitoring point was located at the top of pier No. 3, and the monitoring index was horizontal displacement. The measured value was 18 mm, the traditional calculated value was 12 mm, and the deviation rate was +50%.

[0095] The monitoring point was located at the GYZ300 support, and the monitoring index was shear strain. The measured value was 0.3, the traditional calculated value was 0.2, and the deviation rate was +50%.

[0096] The monitoring point was located at the mid-span of the bridge, and the monitoring index was the vertical vibration frequency. The measured value was 3.5Hz, while the traditional calculated value was 3.0Hz, with a deviation rate of +17%.

[0097] Through the multiphysics coupling simulation steps described above, the stress simulation of the entire system of "water flow-bridge pier-foundation-foundation" was realized for the first time, solving the deviation problem of traditional single physics calculation. The accuracy of structural response data was improved by more than 50%, providing "precise data support" for subsequent early warning.

[0098] Traditional early warning systems only provide "over-limit alarms" (such as alarms triggered when stress exceeds specifications), lack a tiered mechanism, and have no corresponding optimization strategies. This prevents designers from quickly locating problems and developing solutions, resulting in low optimization efficiency. Therefore:

[0099] S4. Based on the structural response data, evaluate the bridge's performance status and generate corresponding warning levels and optimization strategies according to a preset hierarchical warning rule base, forming a warning report. The hierarchical warning rule base includes multiple levels from suggestive warnings to catastrophic warnings. Specifically, generating warning levels and optimization strategies based on the preset hierarchical warning rule base in step S4 includes: comparing the simulated structural response data with preset safety thresholds; if key indicators exceed the safety thresholds, triggering corresponding warning levels based on the extent and degree of exceedance; the optimization strategies are generated from a preset strategy library based on the warning levels, and include adjustments to design parameters, changes to structural form, and reselection of bridge locations; the warning levels include: Level 1 suggestive warning, indicating that the indicator is approaching the threshold and attention is recommended; Level 2 general warning, indicating that the response of local components exceeds limits and design optimization is recommended; Level 3 severe warning, indicating that key components are at risk of damage and major design changes are recommended; Level 4 catastrophic warning, indicating that the structure is at risk of instability and collapse and the current plan should be rejected. Specific operations are as follows:

[0100] First, determine the safety threshold:

[0101] The determination of the safety threshold needs to consider "standard requirements + full life cycle degradation + full-scale test verification" to ensure the scientific validity and safety of the threshold. Specific steps include: First, determining the basic threshold, referring to the "Railway Bridge and Culvert Design Code" TB10002-2017, to determine the basic threshold (e.g., the standard value of tensile stress at mid-span of a T-beam is 1.89 MPa); Second, determining the degradation reduction, considering the degradation of material properties over a 100-year service life (concrete compressive strength degrades by 0.05% annually, tensile strength by 0.08% annually). The yield strength of the steel reinforcement degrades by 0.02% per year. The threshold value after degradation is calculated (tensile stress at mid-span of the T-beam: 1.89MPa × (1 - 0.08 × 100%) = 1.75MPa). Finally, experimental verification is conducted. Ten sets of full-scale model tests of 32m simply supported T-beams (simulating 100 years of load degradation) are carried out at a bridge testing center of China Railway. The measured limit value of the tensile stress at mid-span of the T-beam is 1.8MPa, and the corrected threshold is 1.75MPa (retaining a 3% safety reserve). The final key indicator thresholds are as follows:

[0102] The monitoring index is the tensile stress at the bottom of the mid-span of the T-beam, the safety threshold is 1.75 MPa, the degradation reduction is based on the 8% degradation of the tensile strength of C50 concrete, and the test verification result is the full-scale test limit value of 1.8 MPa;

[0103] The monitoring index is the horizontal displacement of the pier top, the safety threshold is 20mm, the degradation reduction is based on the bridge overturning safety factor of 1.5, and the test verification result is that the critical displacement of the overturning test is 25mm.

[0104] The monitoring index is the shear strain of the bearing rubber pad, the safety threshold is 0.4, the degradation reduction is based on the rubber fatigue life of 1 million cycles, and the test verification result is the critical strain of fatigue test is 0.45.

[0105] The monitoring index is the vertical displacement at mid-span, the safety threshold is 64mm, the degradation reduction is based on the standard L / 600×1.2 dynamic amplification factor, and the test verification result is that the maximum allowable displacement of the vibration test is 65mm.

[0106] Secondly, the alert level is triggered as follows:

[0107] The simulation results of step S3 are compared with the threshold, and a level 4 warning is triggered according to the "exceeding ratio". The specific logic is as follows:

[0108] Level 1 Warning: The indicator is at 80%-100% of the threshold (e.g., tensile stress at mid-span of T-beam is 1.4-1.75MPa). Because the indicator is close to the threshold, there is a potential risk. Therefore, designers are advised to focus on monitoring this part and pay attention to it in subsequent iterations. This will help to avoid the hidden risk of "slowly exceeding the limit" and prevent major changes in the future.

[0109] Level 2 General Warning: When the index of a single local component exceeds the threshold by less than 10% (e.g., tensile stress of 1.75-1.925 MPa at mid-span of a T-beam), it is because the response of the local component exceeds the limit, affecting local performance. Therefore, a local optimization strategy (e.g., adjusting reinforcement parameters) is required. The problem can be solved by local modification, thereby reducing design costs.

[0110] Level 3 Severity Warning: Key component indicators exceed the threshold by 10%-30% (e.g., horizontal displacement of the pier top by 20-26mm). This is because key components are at risk of damage, threatening overall safety. Therefore, a major design change strategy (e.g., increasing the pier cross-section) is required to avoid safety hazards to the entire bridge caused by damage to key components.

[0111] Level 4 Catastrophic Warning: The overall structural indicators exceed the threshold by more than 30% (e.g., mid-span displacement > 83.2 mm). Due to the risk of structural instability and collapse, the current plan is rejected, and a new site is selected or the structural form is changed to fundamentally avoid the risk of collapse and ensure safety throughout the entire life cycle.

[0112] Then, the matching optimization strategy:

[0113] Establish a database linking "early warning level - indicator type - strategy parameters" (storing 200+ strategies). Strategies are formulated based on "mechanical calculations + engineering cases" to ensure feasibility. An example is shown below:

[0114] The warning level is Level II. The risk location is the No. 2 span T-beam. The optimization strategy is to increase the cross-section of the tensile reinforcement. The strategy parameters are: reinforcement diameter from 25mm to 28mm and spacing from 150mm to 120mm. The expected effect is that the tensile stress will increase from 3.2MPa to 1.6MPa (<1.75MPa).

[0115] The warning level is Level 3, the risk location is pier No. 3, the optimization strategy is to change the pier cross section + add anti-slide piles, the strategy parameters are circular pier → rectangular pier (1.5m×1.5m), 3 anti-slide piles (8m long, 0.8m in diameter), the expected effect is horizontal displacement from 18mm → 15mm (<20mm).

[0116] The warning level is Level IV. The risk area is the overall structure. The optimization strategy is to relocate the bridge site and change the structural form. The strategy parameters are to relocate 500m upstream and change from a simply supported beam to a continuous rigid frame. The expected effect is that the mid-span displacement will decrease from 85mm to 55mm (<64mm).

[0117] Finally, an early warning report is generated:

[0118] The early warning report includes "early warning level, risk location (with 3D coordinates), risk cause, optimization strategy, and data comparison table," for example: "Early warning level: Level II general early warning (1 item), Level III severe early warning (1 item); Risk location: Mid-span of T-beam No. 2 (X=120m, Y=30m, Z=5m), top of pier No. 3 (X=152m, Y=30m, Z=8m); Risk cause: Coupled by extreme loads of '35t axle load train + 35m / s strong wind', resulting in excessive tensile stress at the mid-span of the T-beam; Optimization strategy: Adjust the reinforcement of T-beam No. 2 to 28mm diameter and 120mm spacing, and change pier No. 3 to a rectangular section and add anti-slip piles." The early warning report also includes stress cloud diagrams and displacement curves of the monitoring points to ensure that designers can intuitively understand the problem.

[0119] Through the above steps, a system of "quantified threshold - hierarchical early warning - precise strategy" was established, which solved the shortcomings of traditional early warning that was "fuzzy and lacked strategy", improved the efficiency of design optimization, and changed the decision-making basis from "experience judgment" to "data-driven".

[0120] Considering that traditional reports only contain "exceeding limits," lack risk location identification, causal analysis, and visualization, designers need to manually interpret them, which is inefficient. Therefore:

[0121] S5. Output an early warning report, which includes the early warning level, risk location, risk cause, and recommended optimization strategies; the specific technical operation details are as follows:

[0122] The report, generated using JasperReports, is in PDF format and includes six core modules: First, an overview of the warnings (level distribution, number of risks); second, risk details (location coordinates, measured values, thresholds, and exceedance percentages); third, causal analysis (load coupling relationships, structural weak points); fourth, optimization strategies (parameter tables, implementation steps); fifth, simulation visualization (stress cloud diagrams, displacement animation screenshots); and sixth, data attachments (original simulation data, threshold calculation process). The warning report is automatically sent to the designer's email address (via SMTP protocol) and supports online viewing and downloading.

[0123] The above steps improve the completeness of the report information and shorten the time designers spend interpreting it.

[0124] Considering that traditional design optimization is a "one-time modification" without iterative verification, it is impossible to ensure that there are no new risks after optimization, which may lead to hidden dangers in later operation. Therefore, in step S6, the bridge design is modified based on the optimization strategy output in step S5, and steps S1 to S5 are repeated for iterative simulation verification. When the simulation results no longer trigger the level 3 severity warning, and all level 2 general warnings are eliminated or their number is below the preset threshold, the design optimization loop is completed. The specific operation is as follows:

[0125] Perform the first iteration:

[0126] First, design modifications were made, adjusting the reinforcement of T-beam No. 2 (diameter 28mm, spacing 120mm) and changing pier No. 3 to a rectangular section (1.5m × 1.5m) according to the early warning strategy. Simultaneously, the digital twin model was updated, modifying the reinforcement parameters and pier section in the geometric model and reassigning material properties. Then, S2-S3 were repeated, keeping the load conditions unchanged, and the simulation was repeated. Finally, the evaluation results showed that the tensile stress at mid-span of the T-beam decreased to 1.8MPa (still slightly exceeding 1.75MPa), and the pier top displacement decreased to 17mm (meeting the standard). One level-two early warning still exists.

[0127] Perform the second iteration:

[0128] First, modifications were made to further reduce the spacing of the reinforcing bars in the tension zone of the T-beam in span 2 to 100mm. Then, simulation was performed again, and the tensile stress at the mid-span of the T-beam was reduced to 1.6MPa (meeting the standard), with no level 3 or higher warnings and 0 level 2 warnings (below the preset threshold of 2 items). Finally, a closed-loop judgment was performed, and if the condition of "no level 3 warnings + level 2 warnings ≤ 2 items" was met, the iteration was stopped, and the current design was determined to be the final scheme.

[0129] The above iterative closed-loop design ensures that the optimization scheme has no omission risks, improves the structural safety redundancy, and reduces maintenance costs in the later operation stage.

[0130] In summary, the innovative aspects of the graded early warning method for railway bridges provided in this embodiment are as follows: It breaks through the limitations of traditional "simplified models" by constructing a digital twin model that integrates geometry, physics, and environment, achieving a full-system mapping of "bridge-environment." This requires integrating data from multiple fields, including design, materials, and geology, and employing collaborative modeling across multiple software programs. Furthermore, it utilizes "historical data + probabilistic models" to generate multi-level continuous load spectra, covering dynamic loads throughout their entire lifecycle. This combines extreme value theory with load coupling, addressing the shortcomings of traditional isolated loads, and requires cross-disciplinary collaboration between mechanics and statistics. It achieves collaborative simulation of multiple physics fields, including "fluid-solid structure-soil-structural dynamics," which traditional single-physics field simulations cannot achieve. Simultaneously, it establishes a threshold quantification method and graded strategy library for "degradation reduction and experimental verification," combining material degradation with full-scale testing to ensure scientifically sound thresholds. Finally, it forms an iterative closed loop of "simulation-early warning-optimization-verification," linking design optimization with simulation verification to avoid the risk of one-time modifications.

[0131] Example 2

[0132] To implement the method of Example 1, therefore, please refer to Figure 2 As shown, the purpose of Embodiment 2 is to provide a graded early warning system for railway bridges. This system includes a "data acquisition and integration module → digital twin model construction module → simulation calculation module → early warning analysis and decision-making module → visualization and interaction module." The modules interact with each other via a RESTful API interface (communication protocol HTTP / HTTPS, data format JSON). The overall architecture adopts a "distributed deployment + centralized management" approach to ensure stability and scalability. Specific module descriptions are as follows:

[0133] The data acquisition and integration module receives bridge design data and multi-source environmental data; it addresses the problems of traditional data "distributed storage, heterogeneous formats, and delayed updates," specifically including:

[0134] High-definition scanner (Artec Eva, 0.1mm accuracy) is used to scan bridge design drawings and BIM models; GNSS receiver (Trimble R10, centimeter-level positioning accuracy) is used to collect geographical coordinate data of the bridge site; ultrasonic water level gauge (RS485-UTG, 0-10m range, ±0.5%FS accuracy) is used to monitor water level at the bridge site; current meter (LS25-1, 0.05-10m / s range) is used to monitor water flow velocity; meteorological monitoring station (Davis Vantage Pro2) is used to monitor wind speed, temperature, rainfall, etc.; server (Dell PowerEdge R750) is used to store multi-source data and run integrated software.

[0135] The data acquisition client (custom-developed based on C language) automatically connects to devices such as scanners and water level gauges, and collects data at a preset frequency (e.g., design data is collected all at once, and environmental data is collected every 5-10 minutes). The data collected by the data acquisition client is then cleaned (outliers are removed, such as invalid data with wind speed > 67m / s), converted (CAD drawings in DWG format and GIS data in SHP format are uniformly converted to JSON), and loaded (stored in a MySQL 8.0 database) using an ETL data integration tool. This achieves "automatic acquisition - unified integration - real-time sharing" of multi-source data, improving data integrity and data acquisition efficiency.

[0136] The digital twin model building module, connected to the data acquisition and integration module, is used to build and update digital twin models of railway bridges. It includes geometric model building software (Autodesk Revit 2023), multiphysics attribute assignment software (ANSYS AIM 2023), basic geographic information integration software (ArcGIS Pro 3.0), and an automatic model update module (custom-developed, based on Python).

[0137] The steps for using the geometry modeling software (Autodesk Revit 2023) are as follows:

[0138] Launch Revit and create a new "Railway Bridge" project template; import the ETL-converted JSON design data (including the dimensions of T-beams and piers); draw the geometric model at a 1:1 scale: first draw the substructure (foundation → piers), then draw the superstructure (T-beams → bridge deck), and finally add supports and reinforcement; save the model as a .rvt file (approximately 500MB / model) and link it to the bridge site's GIS coordinates (118°25′E, 30°12′N).

[0139] The operation steps of the multiphysics property assignment software (ANSYS AIM 2023) are as follows:

[0140] Import Revit .rvt models and automatically identify component types (T-beams, piers, reinforcement); batch assign material properties according to component type (retrieving material parameter tables from a MySQL database, such as the elastic modulus of C50 concrete being 3.45 × 10⁻⁶). 4 (MPa); Set boundary conditions: Select the bottom of the pier, add "elastic support", and enter a spring stiffness of 1.2 × 10⁻⁶. 5 N / m (retrieved from the experimental data in step S1); save as .aim format for subsequent simulation;

[0141] The operation steps for the basic geographic information integration software (ArcGIS Pro 3.0) are as follows:

[0142] Import the bridge site GIS data (DEM, stratigraphic distribution); import the ANSYS AIM model file, and use the "Coordinate Calibration" function to overlay the bridge model with the GIS data (ensuring the bridge piers are located above the pebble layer); add the "Topographic Profile" function to view the positional relationship between the bridge foundation and the strata; save as a .project file to complete the geographic information integration;

[0143] The operation steps of the model automatic update module (custom-developed, based on Python) are as follows:

[0144] Configure data update trigger conditions (e.g., triggered when the MySQL database design_data table is updated); the module automatically compares the old and new data to identify the changed parts (e.g., the diameter of the bridge pier changes from 1.2m to 1.5m); calls the Revit API to automatically modify the corresponding component parameters and reassign material properties; generates an update log to record the changes and time.

[0145] Through the collaborative work of the aforementioned software, the "automatic construction - dynamic update - multi-attribute fusion" of digital twin models is achieved, which shortens the modeling time and improves the efficiency of model updates.

[0146] The simulation calculation module, connected to the digital twin model construction module, is used to perform multiphysics coupled simulation calculations. It includes multiple computing nodes (performing parallel simulation calculations), a single cluster management node (scheduling computing resources and monitoring task progress), and a distributed storage system (storing simulation result data). The multiphysics coupled simulation calculations include fluid-structure interaction, soil-structure interaction, and structural response calculations. The specific operation process is as follows:

[0147] Fluid-structure interaction operation steps:

[0148] Import the fluid domain and bridge pier models from the digital twin model using ANSYS CFX 2023; set fluid parameters (water density, viscosity) and boundary conditions (inlet velocity, outlet pressure); submit the task to multiple computing nodes for parallel computation; and output the hydrodynamic pressure distribution results (.res format).

[0149] Steps for handling soil-knot interactions:

[0150] By using FLAC3D 7.0 in conjunction with ANSYS Workbench, a foundation model is built in FLAC3D and the stratum parameters are assigned; the foundation model is imported into ANSYS and coupled with the bridge model; settlement loads are applied and submitted to multiple calculation nodes; and the results of foundation reaction force and pier stress are output.

[0151] Structural response operation steps:

[0152] Import the coupled model using ANSYS Mechanical 2023, set monitoring points (20 key locations), load multi-level load spectra, submit to multiple calculation nodes to calculate stress, displacement, and vibration response, and output time-domain / frequency-domain results (.rst format) for each monitoring point.

[0153] A single cluster management node includes task scheduling software (such as LSF 10.1), and the operation steps are as follows:

[0154] Install LSF on the management node and configure multiple (e.g., 5) compute nodes as a compute cluster; create a simulation task queue and allocate nodes according to task complexity (e.g., allocate 3 nodes for extreme load level simulation); set task priorities (optimization iteration tasks have higher priority than the first simulation); monitor task progress in real time and send email notifications (task completion / failure).

[0155] The distributed storage system includes result storage software (GlusterFS 9.0), and the operation steps are as follows:

[0156] Deploy GlusterFS across multiple compute nodes to build a distributed file system; create a "Simulation Results" volume and set data redundancy (replica count 2); automatically store simulation results to the corresponding directory; support retrieving results by time and load level;

[0157] By employing clustered parallel computing, the simulation time for extreme load levels was shortened, the simulation accuracy was improved, and the efficiency of storing and retrieving simulation results was enhanced.

[0158] The early warning analysis and decision-making module, connected to the simulation calculation module, contains the aforementioned hierarchical early warning rule base, used to analyze simulation results and generate early warning reports; wherein:

[0159] The tiered early warning rule base is built on MySQL 8.0 and includes an early warning level table to store the definitions of different levels of early warning; an indicator threshold table to store the thresholds of key indicators; and a strategy matching table to store the association between early warnings and strategies.

[0160] The simulation calculation module is developed based on Java, and the operation steps are as follows:

[0161] Obtain monitoring point result data from the simulation calculation module (e.g., tensile stress at mid-span of T-beam in span 2, 1.85 MPa); query the index threshold table to obtain the corresponding threshold (1.75 MPa); calculate the excess ratio ((1.85 - 1.75) / 1.75 × 100% = 5.7%); match the warning level table to determine it as a level 2 warning; query the strategy matching table to obtain the optimization strategy (reinforcement adjustment); generate a warning analysis report (JSON format).

[0162] Early warning reports are generated using early warning report generation software (such as JasperReports 6.20). The steps are as follows:

[0163] Import JSON data output from the simulation calculation module; call the preset report template (including early warning overview, risk details, strategy table, and visualization charts); generate a PDF report, automatically insert stress cloud diagrams and displacement curves (extracted from simulation results); send the report to the designer's email address (e.g., design@railway.com) via SMTP protocol, and simultaneously store it in the database server;

[0164] The aforementioned software enables "automation, rule-based approach, and traceability" in early warning analysis, shortening the early warning report generation time and improving the accuracy of strategy matching.

[0165] The visualization and interaction module, connected to the early warning analysis and decision-making module, is used to display the simulation process, early warning information, and optimization strategies. This includes immersive large-screen display systems (such as Leyard LED splicing screens, used to display digital twin models, simulation processes, and early warning information in 3D), touch-screen interactive terminals (such as Microsoft Surface Hub 2S, used by designers to manually operate models (rotate, section) and mark risk points), and graphics rendering workstations (such as HP Z8 Fury G5, used to drive real-time rendering of 3D models and process interactive commands).

[0166] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A hierarchical warning method for railway bridges, characterized in that, The method comprises the following steps: S1, constructing a digital twin model of a railway bridge; the digital twin model is a virtual model mapping the whole life cycle state of the physical bridge; S2, obtaining multi-source environment data of a target bridge site area, and generating multi-level load working conditions based on the multi-source environment data; the multi-source environment data comprises geographic and geological data, hydrological data and meteorological data; Wherein, generating multi-level load working conditions specifically comprises: Based on historical data and a probability model, a continuous load spectrum from a normal use state to an extreme adverse event is generated; The load spectrum comprises river scouring and foundation settlement working conditions for evaluating the substructure, wind load and temperature load working conditions for evaluating the superstructure, and train dynamic load working conditions for evaluating the overall structure; S3, loading the multi-level load working conditions to the digital twin model for multi-physical field coupling simulation calculation to obtain structural response data of the bridge at a multi-scale level; the multi-scale level comprises the bridge as a whole, the superstructure, the substructure and key components of the support; Wherein, performing multi-physical field coupling simulation calculation specifically comprises: Performing fluid-structure coupling simulation to simulate the interaction between water flow and pier structure; Performing soil interaction simulation to simulate the interaction between the foundation and the bridge foundation; Calculating the stress, strain, displacement and vibration response of each scale level of the bridge under the multi-level load working conditions; S4, evaluating the performance state of the bridge according to the structural response data, and generating corresponding warning levels and optimization strategies according to a preset hierarchical warning rule library to form a warning report; the hierarchical warning rule library comprises multiple levels from a prompt warning to a disastrous warning; The warning levels comprise: A first level of prompt warning, indicating that the index is close to the threshold value, and suggesting attention; A second level of general warning, indicating that the local component response exceeds the limit, and suggesting optimization design; A third level of serious warning, indicating that the key component has damage risk, and suggesting major design changes; A fourth level of disastrous warning, indicating that the structure has the risk of instability and collapse, and suggesting rejection of the current scheme; S5, outputting the warning report, which comprises the warning level, risk position, risk cause and recommended optimization strategy; S6, modifying the bridge design based on the optimization strategy output in step S5, and repeating steps S1 to S5 for iterative simulation verification; when the simulation result no longer triggers the third level of serious warning, and all second level of general warnings are eliminated or the number is lower than a preset threshold, the design optimization closed loop is completed.

2. The hierarchical warning method for railway bridges according to claim 1, characterized in that, In the step S1 of constructing the digital twin model of the railway bridge, specifically comprising: Based on the bridge design drawings, a geometric model of the bridge is established; Material properties, boundary conditions and connection relationships are given to the geometric model to form a structure calculation model containing multi-physical field properties; Integrate the basic geographic information model of the target bridge site area to complete the construction of the digital twin model.

3. The hierarchical warning method for railway bridges according to claim 1, characterized in that, In the step S4, the preset hierarchical warning rule library is used to generate the warning level and the optimization strategy, specifically comprising: Comparing the structural response data obtained by simulation with the preset safety threshold; If the key index exceeds the safety threshold, the corresponding warning level is triggered according to the exceeding range and degree; The optimization strategy is generated by matching from a preset strategy library based on the early warning level, and the strategy content includes design parameter adjustment, structure form change and bridge site reselection.

4. A hierarchical warning system for a railway bridge, which is used to implement the hierarchical warning method for a railway bridge according to any one of claims 1 to 3, characterized in that, The method comprises the following steps: a data acquisition and integration module for receiving bridge design data and multi-source environmental data; a digital twin model construction module connected with the data acquisition and integration module, for constructing and updating the digital twin model of the railway bridge; a simulation calculation module connected with the digital twin model construction module, for performing multi-physical field coupling simulation calculation; an early warning analysis and decision module connected with the simulation calculation module, which is internally provided with the hierarchical early warning rule library, for analyzing the simulation results and generating an early warning report; a visual interaction module connected with the early warning analysis and decision module, for displaying the simulation process, early warning information and optimization strategy.

Images