Life cycle durability evaluation system of cross-sea bridge in coastal severe corrosion environment

Abstract

This invention relates to the field of facility durability assessment technology, and in particular to a life-cycle durability assessment system for cross-sea bridges in highly corrosive coastal environments. The system includes: a multi-source heterogeneous data acquisition module, used to acquire corrosive environment data, structural response data, component apparent damage data, and historical material performance data through a sensor network, image acquisition equipment, and database deployed on the bridge body and its environment; a data fusion and spatiotemporal registration module, connected to the multi-source heterogeneous data acquisition module, used to clean, align, and standardize the acquired heterogeneous data, and establish a mapping relationship between environmental exposure, material damage, and structural response; and a full-link intelligent system encompassing data perception, fusion analysis, AI prediction, dynamic calibration, and assessment decision-making. Through the fusion of physical mechanisms and data-driven approaches, this system achieves a fundamental shift in cross-sea bridge durability assessment from experience-based judgment to accurate prediction, and from passive maintenance to proactive operation and maintenance.

Description

Technical Field

[0001] This invention relates to the field of facility durability assessment technology, specifically a life-cycle durability assessment system for cross-sea bridges in highly corrosive coastal environments. Background Technology

[0002] As a key hub connecting the transportation network of coastal areas, the construction and operation of cross-sea bridges in highly corrosive coastal environments are of great strategic significance to the country's economic and social development. These large-scale bridge projects are usually designed for a reference period of decades or even a century, and during their service life, they need to withstand the coupled effects of multiple corrosive environments such as the marine atmosphere and splash zones. The structural system of cross-sea bridges is complex, consisting of multiple key components such as pile foundations, abutments, piers, and main beams. The corrosive environments of each component are significantly different, and their durability degradation process exhibits obvious spatiotemporal variability. According to CN116628891A, a digital twin-driven method for assessing the safety of truss structures in coastal environments is disclosed. This technology includes the following steps: Step S1, establishing an integrated profile model, i.e., establishing a multi-scale refined model of the truss structure in a coastal environment based on data monitored by multi-source sensors located at the truss; Step S2, establishing a reduced-order proxy model, i.e., filling in the missing structural response data during monitoring based on the model data from Step S1; Step S3, establishing a safety status assessment model, directly calculating the failure probability of the structure by combining the twin data retrieved in Step S2; Step S4, establishing a real-time dynamic visualization platform, which vividly presents the structural change characteristics and performance evolution trends of the truss based on the data from Steps S1 to S3. This method has the technical effects of "maximizing the use of limited monitoring data, intuitively and in real-time reflecting changes in the truss structure, and accurately and quantitatively assessing the state of the truss structure." In existing methods for assessing the durability of cross-sea bridges, the lack of systematic integration and collaborative analysis of multi-source heterogeneous data, such as corrosion environment, structural response, and material damage, leads to significant data silos in the assessment process. Various monitoring data are often collected and processed independently, making it difficult to establish an accurate spatiotemporal correspondence between environmental exposure information and structural damage status. This prevents the assessment model from fully reflecting the actual corrosion degradation process under the coupled effects of multiple factors. This data-level fragmentation directly causes a deviation between durability assessment and the actual structural state, making it difficult to form a complete analytical chain from environmental excitation to material degradation and then to the evolution of structural performance, thus affecting the accuracy and reliability of the assessment results. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a full life-cycle durability assessment system for cross-sea bridges in highly corrosive coastal environments. It constructs a full-link intelligent system encompassing data perception, fusion analysis, AI prediction, dynamic calibration, and assessment decision-making. Through the integration of physical mechanisms and data-driven approaches, it achieves a fundamental transformation in cross-sea bridge durability assessment, moving from experience-based judgment to accurate prediction, and from passive maintenance to proactive operation and maintenance.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a full life-cycle durability assessment system for cross-sea bridges in highly corrosive coastal environments, comprising: The multi-source heterogeneous data acquisition module is used to acquire corrosive environment data, structural response data, component apparent damage data, and historical material performance data through sensor networks, image acquisition equipment, and databases deployed on the bridge body and in the environment. The data fusion and spatiotemporal registration module is connected to the multi-source heterogeneous data acquisition module. It is used to clean, align and standardize the acquired heterogeneous data, and establish a mapping relationship between environmental exposure, material damage and structural response. The core durability analysis engine is connected to the data fusion and spatiotemporal registration module, including a high-fidelity multiphysics coupling model for fine simulation of structural durability degradation, and a lightweight artificial intelligence agent model as its fast computation agent. The model dynamic update and collaborative calibration module is connected to the multi-source heterogeneous data acquisition module and the core durability analysis engine, respectively. It is used to compare the latest monitoring data with the AI ​​agent model prediction value, and trigger the high-fidelity model to calibrate the parameters of the AI ​​agent model when the deviation exceeds the limit. The life cycle assessment and prediction module is connected to the calibrated core durability analysis engine to calculate the time-varying reliability index and remaining service life of the component. The visualization early warning and decision support module is connected to the full life cycle assessment and prediction module to display results, trigger early warnings, and generate maintenance decision recommendations.

[0005] Preferably, the multi-source heterogeneous data acquisition module includes: The environmental data acquisition unit includes chloride ion concentration sensors, temperature and humidity sensors, wind speed and direction sensors, and surge pressure sensors distributed at different elevations and locations on the bridge. The structural response data acquisition unit includes strain sensors and acceleration sensors arranged on the main beam, cable tower and pier; The apparent damage image acquisition unit includes a high-definition camera deployed at a fixed location and an automatic inspection system for drones equipped with an infrared thermal imager and a lidar, used to acquire images and three-dimensional point cloud data of cracks and corrosion products on the concrete surface. A historical database of material properties is used to integrate, store, and manage the original design parameters, construction process data, and records of all maintenance and inspections for each component.

[0006] Preferably, the high-fidelity multiphysics coupling model specifically includes: The chloride ion transport sub-model is a time-varying diffusion-convection model that considers the coupled effects of internal temperature and humidity and the flow conduction effect of cracks in concrete. The electrochemical corrosion sub-model takes the chloride ion concentration on the steel bar surface and the relative humidity of the environment calculated by the chloride ion transport sub-model as input, and outputs the instantaneous corrosion rate of the steel bar. The structural performance degradation sub-model converts the amount of steel corrosion into the effective cross-sectional loss rate of steel and the cracking history of the concrete cover, and calculates the degradation curve of the component's bearing capacity accordingly.

[0007] Preferably, the artificial intelligence agent model is a spatiotemporal convolutional neural network trained with sample data generated by a high-fidelity multiphysics coupling model, and a constraint term based on the physical-driven corrosion diffusion equation is introduced into the loss function; the input of the model includes at least environmental temperature and humidity time series data, chloride ion concentration on the concrete surface and the current stress level of the component, and the output is the chloride ion concentration distribution at the key section, the steel corrosion depth and the remaining coefficient of the component's bearing capacity.

[0008] Preferably, the process of the model dynamic update and collaborative calibration module calibrating the artificial intelligence agent model is as follows: using the chloride ion concentration value measured by on-site sampling or the crack width obtained by image recognition quantization as a benchmark, the residual between the measured value and the predicted value is calculated; when the residual exceeds the threshold, a high-fidelity multiphysics coupling model is called to perform high-precision simulation, and the simulation result is used as the true value to incrementally train the artificial intelligence agent model using the backpropagation algorithm.

[0009] Preferably, the full life cycle assessment and prediction module constructs a bridge information model integrating data from the entire design, construction, and operation phases, calls a calibrated artificial intelligence proxy model, inputs the predicted future environmental load sequence, simulates the long-term evolution of component durability indices, and uses the first second-order moment method to calculate the time-varying reliability index β(t) of the component under the combined action of corrosion and operational loads. The point at which β(t) first falls below the target reliability index β0 is defined as the theoretical remaining service life of the component.

[0010] Preferably, the visualization early warning and decision support module includes: The early warning unit has preset multi-level durability index thresholds related to the importance of the component. When the predicted index reaches the threshold, it automatically triggers the corresponding level of visual and audible alarm information. The decision support unit has a built-in efficiency-cost optimization algorithm that simulates the impact of different maintenance timings and measures on the remaining service life and total life cycle cost of components, and outputs a decision report containing recommended solutions and quantitative evidence.

[0011] Preferably, the efficiency-cost optimization algorithm in the decision support unit is a multi-objective optimization algorithm used to generate a Pareto optimal solution set to support decision-making.

[0012] This invention provides a life-cycle durability assessment system for cross-sea bridges in highly corrosive coastal environments. Compared with existing technologies, it has the following advantages: 1. By constructing a dual-core analysis engine combining a physical mechanism model and an AI proxy model, the system deeply integrates high-fidelity simulation based on first principles with high-speed prediction driven by data. This enables refined and quantitative simulation of the entire process of chloride ion transport, steel corrosion, and load-bearing capacity degradation. This not only improves the spatiotemporal accuracy of corrosion prediction, but more importantly, it forms an adaptive closed loop of monitoring-prediction-calibration through a dynamic calibration module. This allows the system to continuously learn and evolve, accurately capturing the long-term time-varying patterns of structural performance and providing unprecedented scientific basis for durability assessment.

[0013] 2. Through the multi-source data acquisition and fusion registration module, environmental, response, damage, and historical data are integrated into an organic whole. Based on this, the system seamlessly connects the assessment results with operation and maintenance decisions: the remaining lifetime and reliability indicators provided by the assessment and prediction module are directly transformed into concrete and quantitative maintenance plans by the decision support unit; through multi-objective optimization of efficiency and cost, a Pareto optimal solution set is generated, allowing managers to make scientific trade-offs under multiple constraints such as safety and economy; this forms a complete intelligent chain from perceived data to generated decisions, improving the initiative and scientific nature of operation and maintenance management.

[0014] 3. By coupling temperature, humidity, and crack effects, the unsaturated transport process of chloride ions was accurately described, and the pathogenesis was fully revealed by extrapolating from microscopic electrochemical corrosion to macroscopic load-bearing capacity degradation. At the same time, the AI ​​proxy model that integrates physical information effectively balances computational efficiency and prediction reliability. Through the deep integration of mechanism and data, the system can accurately assess the current state and reliably predict the performance evolution over the next few decades, providing technical support for formulating economical and reasonable full-life maintenance strategies and ensuring the long-term safe service of cross-sea bridges. Attached Figure Description

[0015] Figure 1 This is a block diagram of the overall system architecture of the present invention.

[0016] In the diagram: 11. Multi-source heterogeneous data acquisition module; 111. Environmental data acquisition unit; 112. Structural response data acquisition unit; 113. Apparent damage image acquisition unit; 114. Material performance history database; 12. Data fusion and spatiotemporal registration module; 13. Core durability analysis engine; 131. High-fidelity multi-physics coupling model; 1311. Chloride ion transport sub-model; 1312. Electrochemical corrosion sub-model; 1313. Structural performance degradation sub-model; 132. Artificial intelligence proxy model; 14. Model dynamic update and collaborative calibration module; 15. Full life cycle assessment and prediction module; 16. Visualized early warning and decision support module; 161. Early warning unit; 162. Decision support unit. Detailed Implementation

[0017] 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 of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0018] Please see Figure 1 This invention provides a technical solution: a life-cycle durability assessment system for cross-sea bridges in highly corrosive coastal environments, comprising: The multi-source heterogeneous data acquisition module 11 is used to acquire corrosive environment data, structural response data, component apparent damage data and material performance historical data through a sensor network, image acquisition equipment and database deployed on the bridge body and the environment. The data fusion and spatiotemporal registration module 12 is connected to the multi-source heterogeneous data acquisition module 11, which is used to clean, align and standardize the acquired heterogeneous data, and establish a mapping relationship between environmental exposure, material damage and structural response. The core durability analysis engine 13 is connected to the data fusion and spatiotemporal registration module 12, including a high-fidelity multiphysics coupling model 131 for fine simulation of structural durability degradation, and a lightweight artificial intelligence agent model 132 as its fast computation agent. The model dynamic update and collaborative calibration module 14 is connected to the multi-source heterogeneous data acquisition module 11 and the core durability analysis engine 13, respectively. It is used to compare the latest monitoring data with the AI ​​agent model prediction value, and trigger the high-fidelity model to calibrate the parameters of the AI ​​agent model when the deviation exceeds the limit. The life cycle assessment and prediction module 15 is connected to the calibrated core durability analysis engine 13 to calculate the time-varying reliability index and remaining service life of the component. The visualization early warning and decision support module 16 is connected to the full life cycle assessment and prediction module 15 to display results, trigger early warnings, and generate maintenance decision recommendations.

[0019] In this implementation scheme, the multi-source heterogeneous data acquisition module 11 comprehensively acquires environmental, response, appearance, and historical data reflecting the structural state through a sensor network and detection equipment distributed throughout the bridge body and environment, forming a multi-dimensional information foundation for durability assessment. The data fusion and spatiotemporal registration module 12 performs spatiotemporal alignment and standardization processing on the multi-source data, establishing a quantitative mapping relationship between environmental exposure parameters, material damage characteristics, and structural mechanical response, solving the problem of fusion and utilization of multi-source heterogeneous data. The core durability analysis engine 13 adopts a dual-model architecture that combines a high-fidelity multi-physics coupling model 131 with a lightweight artificial intelligence proxy model 132, maintaining the simulation accuracy of the physical process while meeting the computational requirements of real-time assessment. Efficiency requirements; The model dynamic update and collaborative calibration module 14 continuously compares monitoring data with prediction results, and triggers the high-fidelity model 131 to calibrate the parameters of the proxy model 132 when the deviation exceeds the limit, forming an intelligent analysis system with self-optimization capabilities; The full life cycle assessment and prediction module 15 calculates the time-varying reliability index of components based on the calibrated analysis results, and realizes the quantitative prediction of the remaining service life; Finally, the visualization early warning and decision support module 16 transforms the complex assessment results into intuitive multi-level early warning signals and quantitative maintenance decision suggestions, completing the closed-loop management from data collection to operation and maintenance decision-making; A full-link durability assessment system from data perception to intelligent decision-making is constructed.

[0020] Specifically, the multi-source heterogeneous data acquisition module 11 includes: The environmental data acquisition unit 111 includes chloride ion concentration sensors, temperature and humidity sensors, wind speed and direction sensors, and surge pressure sensors distributed at different elevations and locations on the bridge. The structural response data acquisition unit 112 includes strain sensors and acceleration sensors arranged on the main beam, cable tower and pier; The apparent damage image acquisition unit 113 includes a high-definition camera deployed at a fixed point and an automatic inspection system for drones equipped with an infrared thermal imager and a lidar, used to acquire images and three-dimensional point cloud data of cracks and corrosion products on the concrete surface. The Material Performance Historical Database 114 is used to integrate, store, and manage the original design parameters, construction process data, and maintenance and inspection records of each component.

[0021] In this embodiment, the environmental data acquisition unit 111 directly monitors key environmental driving factors that lead to steel corrosion and concrete deterioration by deploying multiple types of sensors at different elevations and key locations, providing input boundary conditions for corrosion process analysis; the structural response data acquisition unit 112 captures the changes in the mechanical state of the structure under the combined action of environmental and operational loads in real time by deploying strain and acceleration sensors at key load-bearing locations, reflecting the actual working state and performance degradation signs of the structure; the apparent damage image acquisition unit 113 comprehensively utilizes fixed camera points and mobile drone inspection systems, combined with visible light, infrared, and lidar. The technology enables the acquisition of multi-scale damage information from two-dimensional appearance images to three-dimensional geometric morphology, accurately quantifying crack development, spalling areas, and the distribution of corrosion products. The 114-system material performance history database integrates static and dynamic data from the entire life cycle, from design and construction to operation and maintenance, providing a complete data chain for durability analysis, including initial state benchmarks and performance degradation processes. Through the collaborative work of the four units, a complete data foundation is formed, encompassing environmental excitation to structural response, material degradation to appearance damage, and current state to historical evolution, providing multi-dimensional and comprehensive evidence support for subsequent fusion analysis and intelligent assessment.

[0022] Specifically, the high-fidelity multiphysics coupling model 131 includes: The chloride ion transport sub-model 1311 is a time-varying diffusion-convection model that considers the coupling effects of internal temperature and humidity and the flow conduction effect of cracks in concrete. The electrochemical corrosion sub-model 1312 takes the chloride ion concentration on the steel surface and the relative humidity of the environment calculated by the chloride ion transport sub-model as input, and outputs the instantaneous corrosion rate of the steel. The structural performance degradation sub-model 1313 converts the amount of steel corrosion into the effective section loss rate of steel and the cracking history of the concrete cover, and calculates the degradation curve of the component's bearing capacity accordingly.

[0023] In this embodiment, the chloride ion transport sub-model 1311 is based on the unsaturated concrete transport theory, comprehensively considering the significant influence of the coupling effect of temperature and humidity on the chloride ion diffusion coefficient, and introducing crack width and distribution parameters to characterize the rapid channel effect of cracks on chloride ion transport, thereby achieving accurate prediction of the spatiotemporal evolution law of chloride ions inside concrete; the electrochemical corrosion sub-model 1312, based on the chloride ion concentration on the steel surface and the relative humidity of the environment calculated by the chloride ion transport sub-model 1311, determines the activation corrosion state and instantaneous corrosion rate of the steel based on the principle of electrochemical kinetics, accurately describing the post-passivation film damage. The corrosion development process of reinforcing steel; the structural performance degradation sub-model 1313, based on the corrosion amount of reinforcing steel output by the electrochemical corrosion sub-model 1312, first calculates the time-varying loss of the effective cross section of the reinforcing steel, then simulates the expansion and cracking process of the concrete cover through the corrosion product expansion theory, and finally, based on the cross section analysis method and material constitutive relation, quantitatively calculates the time-varying degradation curves of the component's bearing capacity and stiffness, fully presenting the intrinsic relationship from material-level damage to component-level performance degradation; through the three sub-models, a complete physical process simulation of transport-corrosion-degradation is formed, providing a clear and complete theoretical basis for durability assessment.

[0024] Specifically, the artificial intelligence agent model 132 is a spatiotemporal convolutional neural network trained with sample data generated by the high-fidelity multiphysics coupling model 131, and a constraint term based on the physical-driven corrosion diffusion equation is introduced into the loss function; the input of the model includes at least environmental temperature and humidity time series data, chloride ion concentration on concrete surface and current stress level of component, and the output is chloride ion concentration distribution at key sections, steel corrosion depth and component bearing capacity remaining coefficient.

[0025] In this embodiment, supervised training is conducted using massive sample data generated by the high-fidelity multiphysics coupling model 131 under various typical environments and initial conditions, ensuring the physical correctness and coverage of the training data. The spatiotemporal convolutional neural network architecture effectively captures and learns the long-term dependencies and non-uniform spatial damage distribution in environmental time-series data, thereby achieving accurate mapping of the spatiotemporal evolution of the corrosion process. By introducing a constraint term based on the physics-driven corrosion diffusion equation into the loss function, and incorporating prior physical knowledge such as Fick's second law as a regularization method into the network training process, the network prediction results are forced to not only fit the data but also... In accordance with basic physical laws, this design effectively overcomes the problem of illogical predictions that may occur in purely data-driven models, and significantly improves the model's generalization ability and extrapolation reliability. The artificial intelligence proxy model 132 receives multi-dimensional inputs such as environmental temperature and humidity time-series data, chloride ion concentration on concrete surface, and current stress level of components. It quickly outputs key durability indicators such as chloride ion concentration distribution at key sections, steel corrosion depth, and residual bearing capacity coefficient of components through forward inference. While maintaining the physical accuracy of the high-fidelity model 131, it achieves an order-of-magnitude improvement in computational efficiency, providing core technical support for real-time durability assessment and long-term prediction.

[0026] Specifically, the process of the model dynamic update and collaborative calibration module 14 calibrating the artificial intelligence agent model 132 is as follows: using the chloride ion concentration value measured by on-site sampling or the crack width obtained by image recognition quantization as a benchmark, the residual between the measured value and the predicted value is calculated; when the residual exceeds the threshold, the high-fidelity multiphysics coupling model 131 is called to perform high-precision simulation, and the simulation result is used as the true value to incrementally train the artificial intelligence agent model 132 using the backpropagation algorithm.

[0027] In this embodiment, the model dynamic update and collaborative calibration module 14 obtains the measured chloride ion concentration value inside the concrete by on-site sampling, or quantifies the crack width on the concrete surface by image recognition technology, and uses these physical quantities that directly reflect the actual state of the structure as verification benchmarks. The system then calculates the residual between these measured values ​​and the corresponding predicted values ​​of the artificial intelligence agent model 132. When the residual exceeds a preset threshold, it indicates that the prediction of the agent model 132 has deviated significantly. At this time, the module automatically triggers the high-fidelity multiphysics coupling model 131 to perform local high-precision simulation at the corresponding spatiotemporal location. The calculation results of the high-fidelity model 131 based on first principles are regarded as the "physical truth value" that is closer to the real value in the current state. Finally, the module uses the simulation results of the high-fidelity model 131 as the training target and uses the backpropagation algorithm to perform a round of targeted incremental training on the network parameters of the artificial intelligence agent model 132. This gradual parameter fine-tuning not only effectively corrects the prediction deviation of the model, but also avoids the computational resource consumption caused by complete retraining, thereby realizing the continuous optimization and self-improvement of the performance of the artificial intelligence agent model 132 in the actual service process, and ensuring the accuracy of long-term prediction.

[0028] Specifically, the full life cycle assessment and prediction module 15 constructs a bridge information model integrating design, construction and operation data, calls the calibrated artificial intelligence agent model 132, inputs the predicted future environmental load sequence, simulates the long-term evolution of component durability index, and uses the first second moment method to calculate the time-varying reliability index β(t) of the component under the combined action of corrosion and operation loads. The time point when β(t) first falls below the target reliability index β0 is defined as the theoretical remaining service life of the component.

[0029] In this embodiment, the full life cycle assessment and prediction module 15 calls the artificial intelligence proxy model 132 optimized by the dynamic update and collaborative calibration module 14, and inputs the future environmental load sequence predicted based on historical data and climate models to efficiently simulate the long-term evolution process of key durability indicators of components under continuous corrosion. Based on the evolution results, the module adopts the first-order second-moment reliability theory, treating durability indicators such as steel corrosion depth and concrete cover cracking state as random variables along with operational loads. By establishing a function and calculating its reliability index β(t), the module quantitatively characterizes the safety margin of the component under the coupled action of time-varying corrosion and random loads. Finally, by monitoring the change curve of the reliability index β(t) over time, the module scientifically defines the time point when it first falls below the preset target reliability index β0 as the theoretical remaining service life of the component. This realizes the transformation from deterministic empirical judgment to probabilistic accurate prediction, providing a key theoretical basis for the preventive maintenance and optimization decision-making of bridges.

[0030] Specifically, the visualization early warning and decision support module 16 includes: The early warning unit 161 has preset multi-level durability index thresholds related to the importance of the component. When the predicted index reaches the threshold, it automatically triggers the corresponding level of visual and audible alarm information. The decision support unit 162 has a built-in energy-cost optimization algorithm that simulates the impact of different maintenance timings and measures on the remaining service life and total life cycle cost of components, and outputs a decision report containing recommended solutions and quantitative basis.

[0031] In this embodiment, the early warning unit 161, based on the structural system reliability theory, presets differentiated multi-level durability index thresholds according to the structural importance of different components such as main beams, cable towers, and piers and their functional positions in the system. When the time-varying reliability index β(t) or other key durability parameters transmitted by the full life cycle assessment and prediction module 15 reaches the corresponding threshold, the unit immediately activates a graded response mechanism from prompts and warnings to severe alarms, conveying accurate early warning information to management personnel through various means such as graphical interface color changes, flashing prompts, and audio-visual signals. The decision support unit 162, based on this, integrates a full life cycle assessment that considers time value. The long-term cost analysis model and performance evaluation algorithm construct a multi-dimensional decision space of "maintenance timing - maintenance measures - performance output - cost input". The system simulates the effect of different maintenance strategy combinations on the extension of the remaining service life of components and the changes in total cost generated throughout the entire life cycle. Finally, a series of recommended schemes that achieve the optimal balance between safety performance and economic cost are generated through a multi-objective optimization algorithm. Each scheme is provided with quantitative decision-making basis, including expected life gain, present value of cost, and degree of risk reduction. This transforms the complex durability assessment results into intuitive and operable management decisions, significantly improving the scientific and economic nature of bridge operation and maintenance management.

[0032] Specifically, the efficiency-cost optimization algorithm in decision support unit 162 is a multi-objective optimization algorithm used to generate Pareto optimal solution sets to support decision-making.

[0033] In this embodiment, the efficiency-cost multi-objective optimization algorithm simulates a large number of possible maintenance strategy combinations, representing each strategy as a solution in a multi-dimensional objective space, which includes at least two conflicting optimization objectives: maximizing maintenance efficiency and minimizing the total lifecycle cost. The algorithm filters these solutions using Pareto dominance relations, ultimately generating a Pareto optimal frontier composed of non-dominated solutions. Each solution on this frontier represents the highest maintenance efficiency achievable under a given cost constraint, or the lowest cost input required under specific efficiency requirements. There are no other solutions that are better across all objectives. The decision support unit 162 presents the manager with this set of Pareto optimal solutions in an optimal balance, rather than a single optimal solution. This allows decision-makers to clearly understand the possible outcomes under different decision preferences, and thus make a comprehensive trade-off based on actual budget constraints, safety requirements, and policy orientation, ultimately selecting the optimal maintenance strategy that best meets current management needs. This process establishes subjective decision-making on the basis of objective quantitative analysis, significantly improving the scientific and systematic nature of bridge operation and maintenance decisions.

[0034] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0035] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A life-cycle durability assessment system for cross-sea bridges in highly corrosive coastal environments, characterized in that: include: The multi-source heterogeneous data acquisition module (11) is used to acquire corrosive environment data, structural response data, component apparent damage data and material performance historical data through a sensor network, image acquisition equipment and database deployed on the bridge body and environment; The data fusion and spatiotemporal registration module (12) is connected to the multi-source heterogeneous data acquisition module (11) to clean, align and standardize the acquired heterogeneous data, and establish a mapping relationship between environmental exposure, material damage and structural response. The core durability analysis engine (13) is connected to the data fusion and spatiotemporal registration module (12), which includes a high-fidelity multiphysics coupling model (131) for fine simulation of structural durability degradation, and a lightweight artificial intelligence agent model (132) as its fast computation agent. The model dynamic update and collaborative calibration module (14) is connected to the multi-source heterogeneous data acquisition module (11) and the core durability analysis engine (13), respectively. It is used to compare the latest monitoring data with the AI ​​agent model prediction value and trigger the high-fidelity model to calibrate the parameters of the AI ​​agent model when the deviation exceeds the limit. The full life cycle assessment and prediction module (15) is connected to the calibrated core durability analysis engine (13) to calculate the time-varying reliability index and remaining service life of the component. The visualization early warning and decision support module (16) is connected to the full life cycle assessment and prediction module (15) to display results, trigger early warnings and generate maintenance decision suggestions.

2. The full life-cycle durability assessment system for cross-sea bridges in highly corrosive coastal environments according to claim 1, characterized in that: The multi-source heterogeneous data acquisition module (11) includes: The environmental data acquisition unit (111) includes chloride ion concentration sensors, temperature and humidity sensors, wind speed and direction sensors and surge pressure sensors distributed at different elevations and locations on the bridge. The structural response data acquisition unit (112) includes strain sensors and acceleration sensors arranged on the main beam, cable tower and pier; The apparent damage image acquisition unit (113) includes a high-definition camera deployed at a fixed point and an unmanned aerial vehicle automatic inspection system equipped with an infrared thermal imager and a lidar, used to acquire images and three-dimensional point cloud data of concrete surface cracks and corrosion products. The historical database of material properties (114) is used to integrate, store and manage the original design parameters, construction process data and maintenance and inspection records of each component.

3. The full life-cycle durability assessment system for cross-sea bridges in highly corrosive coastal environments according to claim 1, characterized in that: The high-fidelity multiphysics coupling model (131) specifically includes: The chloride ion transport sub-model (1311) is a time-varying diffusion-convection model that considers the coupling effects of internal temperature and humidity and the flow conduction effect of cracks in concrete. The electrochemical corrosion sub-model (1312) takes the chloride ion concentration on the steel surface and the relative humidity of the environment calculated by the chloride ion transport sub-model as input, and outputs the instantaneous corrosion rate of the steel. The structural performance degradation sub-model (1313) transforms the amount of steel corrosion into the effective cross-sectional loss rate of steel and the cracking history of the concrete cover, and calculates the degradation curve of the component's bearing capacity accordingly.

4. The full life-cycle durability assessment system for cross-sea bridges in highly corrosive coastal environments according to claim 1, characterized in that: The artificial intelligence agent model (132) is a spatiotemporal convolutional neural network trained with sample data generated by the high-fidelity multiphysics coupling model (131), and a constraint term based on the physical-driven corrosion diffusion equation is introduced into the loss function; the input of the model includes at least environmental temperature and humidity time series data, chloride ion concentration on concrete surface and current stress level of component, and the output is chloride ion concentration distribution at key sections, steel corrosion depth and component bearing capacity remaining coefficient.

5. The full life-cycle durability assessment system for cross-sea bridges in highly corrosive coastal environments according to claim 1, characterized in that: The process of the model dynamic update and collaborative calibration module (14) calibrating the artificial intelligence agent model (132) is as follows: using the chloride ion concentration value measured by on-site sampling or the crack width obtained by image recognition quantification as a benchmark, the residual between the measured value and the predicted value is calculated. When the residual exceeds the threshold, the high-fidelity multiphysics coupling model (131) is called to perform high-precision simulation, and the simulation result is used as the true value. The backpropagation algorithm is used to incrementally train the artificial intelligence agent model (132).

6. The full life-cycle durability assessment system for cross-sea bridges in highly corrosive coastal environments according to claim 1, characterized in that: The full life cycle assessment and prediction module (15) constructs a bridge information model integrating design, construction and operation data, calls the calibrated artificial intelligence agent model (132), inputs the predicted future environmental load sequence, simulates the long-term evolution of component durability index, and uses the first second moment method to calculate the time-varying reliability index β(t) of the component under the combined action of corrosion and operation loads. The time point when β(t) first falls below the target reliability index β0 is defined as the theoretical remaining service life of the component.

7. The full life-cycle durability assessment system for cross-sea bridges in highly corrosive coastal environments according to claim 1, characterized in that: The visualization early warning and decision support module (16) includes: The early warning unit (161) has preset multi-level durability index thresholds related to the importance of the component. When the predicted index reaches the threshold, it automatically triggers visual and audible alarm information of the corresponding level. The decision support unit (162) has a built-in efficiency-cost optimization algorithm to simulate the impact of different maintenance timings and measures on the remaining service life and total life cycle cost of components, and outputs a decision report containing recommended solutions and quantitative basis.

8. The full life-cycle durability assessment system for cross-sea bridges in highly corrosive coastal environments according to claim 7, characterized in that: The efficiency-cost optimization algorithm in the decision support unit (162) is a multi-objective optimization algorithm used to generate Pareto optimal solution sets to support decision-making.

Images