Degradation prediction-based concrete bridge preventive maintenance decision-making method

Abstract

The invention discloses a concrete bridge preventive maintenance decision-making method based on degradation prediction, and relates to the technical field of bridge degradation prediction maintenance, and the method comprises the steps: collecting the degradation characteristic data of a concrete bridge through a bridge detection device; according to the method, various types of degradation characteristic data are collected through the bridge detection equipment, the standardized bridge state characteristic vector set is generated, the data-mechanism fusion inversion model is adopted, the actually collected data and the inherent mechanism of bridge degradation are combined, the estimation result is made to better conform to the actual situation, and the estimation accuracy is improved. Dynamic data assimilation is carried out according to a state space model, a probabilistic estimation set of a degradation driving state is input into a predetermined bridge degradation mechanism model for long-term deterministic simulation prediction to generate a plurality of future performance degradation paths, and all the paths form a degradation track cloud picture reflecting prediction uncertainty; the optimal maintenance strategy set can reduce the expected life cycle cost to the maximum extent on the premise that the bridge maintenance requirement is met.

Description

Technical Field

[0001] This invention relates to the field of Dendrobium officinale processing technology, and more particularly to a preventive maintenance decision-making method for concrete bridges based on degradation prediction. Background Technology

[0002] As a key component of transportation infrastructure, concrete bridges play an indispensable role in national economic and social development. Their safety and durability are directly related to the smooth flow of transportation, the safety of people's lives and property, and the stable development of the social economy. With the increase of service life, concrete bridges will inevitably experience various degradation phenomena under the long-term effects of natural environment (such as humidity changes, chloride ion corrosion, etc.) and traffic loads, such as concrete cracking and steel corrosion. These problems will gradually reduce the load-bearing capacity and service performance of the bridge, and in severe cases, may even lead to bridge collapse accidents.

[0003] Currently, preventive maintenance methods based on degradation prediction still have significant limitations in research and practice: existing methods mostly rely on observable damage data on the bridge surface, which are lagging reflections of degradation results. Although periodic inspections can obtain some basic information about the bridge, the fixed inspection cycle often fails to capture dynamic changes in the bridge's condition in a timely manner, making it difficult to achieve real-time monitoring and early warning of bridge degradation. They lack a continuous description of the bridge degradation process, cannot comprehensively and accurately reflect the actual condition of the bridge, and cannot directly reveal the key states of degradation driven by internal concrete humidity and chloride ion concentration. Prediction models are based on indirect correlations, resulting in insufficient accuracy and reliability. Most prediction models are deterministic models or statistical models based on historical data, and their outputs are mostly single remaining life point estimates. They fail to fully quantify the comprehensive prediction uncertainty caused by material performance variations, environmental uncertainties, and model errors, making decision-making based on this lack a risk measurement basis. To address the aforementioned technical shortcomings, a solution is proposed. Summary of the Invention

[0004] The purpose of this invention is to collect and process various types of bridge degradation characteristic data, use a data-mechanism fusion inversion model to estimate the degradation driving state inside the bridge, perform long-term deterministic simulation to predict the future performance degradation trajectory of the bridge, construct a stochastic optimization model to solve for the optimal maintenance strategy set, formulate a detailed maintenance plan based on the optimal maintenance strategy set, and dynamically adjust and optimize the maintenance strategies.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a decision-making method for preventive maintenance of concrete bridges based on degradation prediction, comprising the following steps: Step 1: Collect degradation characteristic data of concrete bridges using bridge inspection equipment. The degradation characteristic data includes time-series data from automated monitoring, discrete data from periodic inspections, and historical environmental data. The degradation characteristic data is then cleaned, spatiotemporally aligned, and standardized to generate a standardized bridge state feature vector set. Step 2: Based on the standardized bridge state feature vector set, the degradation driving state inside the bridge is estimated through the data-mechanism fusion inversion model. The standardized bridge state feature vector set is used as input features, and the humidity field and chloride ion concentration field inside the concrete are used as state variables. Dynamic data assimilation is performed according to the state space model to obtain the probabilistic estimate set of the degradation driving state at the current moment. Step 3: Input each sample in the probabilistic estimation set of degradation driving states into the predetermined bridge degradation mechanism model, perform long-term deterministic simulation to predict the future performance degradation trajectory of the bridge, generate multiple future performance degradation paths, and integrate all paths to form a degradation trajectory cloud map that reflects the uncertainty of the prediction. Step 4: From the probability distribution of future performance degradation trajectories in the degradation trajectory cloud map, construct a stochastic optimization model with conditionally triggered maintenance rules as decision variables and minimizing expected lifecycle costs as a multi-objective, and output the optimal maintenance strategy set by solving the model. Step 5: Develop a detailed maintenance plan based on the optimal maintenance strategy set, specifying the maintenance time, maintenance measures and maintenance scope. During the implementation of the maintenance plan, continuously adjust and optimize the maintenance strategies dynamically.

[0006] Furthermore, a standardized bridge state feature vector set is generated, the specific process of which is as follows: S100: Automatically acquire continuous time-series data from the deployed sensor network at a fixed frequency, and label each data point with a precise timestamp and sensor spatial location coordinates. Integrate the acquired discrete data, and associate each data point with the detection date, detection point number, and the component number to which it belongs. S102. Create a mapping table that associates sensor ID, detection point, component number with bridge, and perform data cleaning and continuous proxy sequence generation. From the cleaned continuous sequence, extract statistical and mechanistic features that can directly and sensitively reflect the process of concrete degradation, including concrete internal and surface temperature and humidity data, chloride ion concentration data, and bridge structure displacement and deformation data. For degradation characteristic data, not only are absolute values ​​recorded, but also the changes compared to the previous detection, the ratio of the historical maximum value, and the mean and standard deviation of the statistical distribution parameters of data from different detection points on the same component are calculated. S103. The standardized feature vectors of all components at the same time point are concatenated with the global environment feature vector to form a complete, standardized, spatiotemporally aligned bridge state feature vector set.

[0007] Furthermore, by running a data-mechanism fusion inversion model, the degradation driving state inside the concrete bridge is estimated. The specific process is as follows: S201: Discretize the pore relative humidity distribution H and free chloride ion concentration distribution C within the concrete cover into a state vector, denoted as X= , as input to the running data-mechanism fusion inversion model; S202: Constructing a state-space model includes a state transition model and an observation model. The physicochemical equations describing the evolution of the internal humidity field and chloride ion concentration field of concrete are used as the state transition model. The mathematical function of the state transition model is constructed to describe the dynamic evolution of the state vector from time k-1 to time k. Based on the mathematical mapping relationship between state variables and the standardized bridge state feature vector set that can be actually obtained, an observation model of the connection state is established. N state vector samples are randomly generated to form the initial state set. ; S203: Based on the initial state set and dynamic data assimilation, the output of the running data-mechanism fusion inversion model is obtained. The output layer is the probabilistic estimate set of the output degradation driving state at the current time.

[0008] Furthermore, dynamic data assimilation is performed based on the state-space model to obtain a probabilistic estimate set of the degradation-driven state at the current moment, specifically including the following: For each individual in the current state set Based on the state transition model, the process proceeds to the next time step to obtain the predicted driving state. ; Each predicted driving state individual The corresponding predicted observation values ​​are calculated using the observation model. The observation vectors obtained after actual new collection and standardization Input, by comparing the actual observed values ​​with the set of all predicted observed values ​​{ The mean and covariance of} are used to calculate the optimal gain coefficient, and this coefficient is then used to correct the individual predicted driving state. , This is called reverse update, based on the observed residuals. Inverse optimization is used to estimate the internal state of an individual. The new set of states obtained after the update { , which is the probabilistic estimate set of the degenerate driving state obtained at time k, i={1,2,……,N}.

[0009] Furthermore, each sample in the probabilistic estimation set of degradation-driven states is input into a predetermined bridge degradation mechanism model to perform long-term deterministic simulations and predict the future performance degradation trajectory of the bridge. The specific process is as follows: S400. Obtain the individual internal control states in the probabilistic estimation set of degradation-driven states, bind them with deterministic future environmental scenario assumptions, and form a complete future evolution input scenario package. Starting from the initial internal state sample, under the drive of the future environmental scenario, run the degradation mechanism model to perform long-term stepwise time accumulation and calculate the evolution of the internal state field at each time point. S401. At each preset output time point in the simulation process, based on the currently simulated internal state, the performance index values ​​calculated at each output time point are connected in chronological order to determine the future performance degradation path. S402. Aggregate the N simulated deterministic future performance degradation paths by time point and calculate the degradation index statistics for each time point; ;

[0010] for Expected value of performance metrics at any given time. For the nth path in Performance values ​​at any given time for The standard deviation of the time-matter performance metric, where N is the number of simulated paths.

[0011] Furthermore, multiple future performance degradation paths are generated, and all paths are integrated to form a degradation trajectory cloud map reflecting the uncertainty of prediction. The specific process is as follows: A degradation trajectory cloud map is constructed with time as the horizontal axis and performance index as the vertical axis. The cloud map is defined by the region of quantile trajectory as the prediction interval, and the mean trajectory and median trajectory can be optionally superimposed as the central trend line to form a future performance degradation trajectory cloud map that intuitively reflects the uncertainty of prediction. Plot the lower bound percentile curve and the upper bound percentile curve, and fill the area between the two percentile curves to form the prediction interval band; Within the prediction interval, draw the mean curve or median curve as the central trend line.

[0012] Furthermore, based on the probability distribution of future performance degradation trajectories in the degradation trajectory cloud map, a stochastic optimization model is constructed with conditionally triggered maintenance rules as decision variables and minimizing the expected lifecycle cost as a multi-objective. The optimal maintenance strategy set is then output by solving the model. The specific process is as follows: S600: Transform maintenance decisions into a dynamic condition-triggered mode, setting the maintenance trigger condition as the degradation index exceeding the warning value; S601. Establish mathematical programming conditions with the single objective of minimizing the expected total life cycle cost, calculate the expected total cost based on the probability distribution of the degradation trajectory, and output a specific set of optimal condition-triggered maintenance strategies defined by the optimal parameters.

[0013] Furthermore, a detailed maintenance plan is developed based on the optimal maintenance strategy set, specifying the maintenance time, maintenance measures, and maintenance scope. During the implementation of the maintenance plan, the maintenance strategies are continuously adjusted and optimized dynamically. Specific steps include: Each strategy is transformed into a correspondence between monitoring and early warning values ​​and maintenance actions. Based on the triggering conditions, the specific scope of each measure is planned in advance. Real-time monitoring and periodic testing data from step one are continuously received and analyzed. The data are compared in real time with the various triggering values ​​preset in the strategy set, and an alarm is automatically generated to produce a maintenance work order. After maintenance is carried out, the latest and most comprehensive monitoring data are input into the data-mechanism fusion inversion model in step two at fixed intervals to re-estimate the internal degradation driving state of the bridge.

[0014] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: This decision-making method for preventive maintenance of concrete bridges based on degradation prediction collects various types of degradation characteristic data through bridge inspection equipment, generating a standardized set of bridge state characteristic vectors. Employing a data-mechanism fusion inversion model, it combines the actually collected data with the inherent mechanisms of bridge degradation. This approach fully utilizes information from actual observation data while adhering to the physical laws of bridge degradation, making the estimation results more consistent with reality. Dynamic data assimilation based on a state-space model accurately estimates the internal degradation driving states of the bridge. The probabilistic estimation set of these degradation driving states is input into a predetermined bridge degradation mechanism model for long-term deterministic simulation and prediction, taking into account various possible degradation scenarios. Multiple future performance degradation paths are generated, and all paths are combined into a degradation trajectory cloud map reflecting the uncertainty of the prediction. This visually demonstrates the possible range and probability distribution of future bridge performance degradation, allowing decision-makers to clearly understand the degree of uncertainty in the prediction results. The optimal maintenance strategy set ensures that, while meeting bridge maintenance requirements, the expected life-cycle cost is minimized, improving the economy and efficiency of bridge maintenance and achieving optimal resource allocation. Attached Figure Description

[0015] Figure 1 A schematic diagram of the overall structure of the method steps of the present invention is shown. Detailed Implementation

[0016] 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.

[0017] Example 1: like Figure 1 As shown, the preventive maintenance decision-making method for concrete bridges based on degradation prediction includes the following steps: Step 1: Collect degradation characteristic data of concrete bridges using bridge inspection equipment. The degradation characteristic data includes time-series data from automated monitoring, discrete data from periodic inspections, and historical environmental data. The degradation characteristic data is then cleaned, spatiotemporally aligned, and standardized to generate a standardized bridge state feature vector set. Step 2: Based on the standardized bridge state feature vector set, the degradation driving state inside the bridge is estimated through the data-mechanism fusion inversion model. The standardized bridge state feature vector set is used as input features, and the humidity field and chloride ion concentration field inside the concrete are used as state variables. Dynamic data assimilation is performed according to the state space model to obtain the probabilistic estimate set of the degradation driving state at the current moment. Step 3: Input each sample in the probabilistic estimation set of degradation driving states into the predetermined bridge degradation mechanism model, perform long-term deterministic simulation to predict the future performance degradation trajectory of the bridge, generate multiple future performance degradation paths, and integrate all paths to form a degradation trajectory cloud map that reflects the uncertainty of the prediction. Step 4: From the probability distribution of future performance degradation trajectories in the degradation trajectory cloud map, construct a stochastic optimization model with conditionally triggered maintenance rules as decision variables and minimizing expected lifecycle costs as a multi-objective, and output the optimal maintenance strategy set by solving the model. Step 5: Develop a detailed maintenance plan based on the optimal maintenance strategy set, specifying the maintenance time, maintenance measures and maintenance scope. During the implementation of the maintenance plan, continuously adjust and optimize the maintenance strategies dynamically.

[0018] The standardized bridge state feature vector set is generated through the following process: S100: Automatically acquire continuous time-series data from the deployed sensor network at a fixed frequency, and label each data point with a precise timestamp and sensor spatial location coordinates. Integrate the acquired discrete data, and associate each data point with the detection date, detection point number, and the component number to which it belongs. S102. Create a mapping table that associates sensor ID, detection point, component number with bridge, and perform data cleaning and continuous proxy sequence generation. From the cleaned continuous sequence, extract statistical and mechanistic features that can directly and sensitively reflect the process of concrete degradation, including concrete internal and surface temperature and humidity data, chloride ion concentration data, and bridge structure displacement and deformation data. For degradation characteristic data, not only are absolute values ​​recorded, but also the changes compared to the previous detection, the ratio of the historical maximum value, and the mean and standard deviation of the statistical distribution parameters of data from different detection points on the same component are calculated. S103. The standardized feature vectors of all components at the same time point are concatenated with the global environment feature vector to form a complete, standardized, spatiotemporally aligned bridge state feature vector set.

[0019] The degradation driving state inside concrete bridges is estimated by running a data-mechanism fusion inversion model. The specific process is as follows: S201: Discretize the pore relative humidity distribution H and free chloride ion concentration distribution C within the concrete cover into a state vector, denoted as X= , as input to the running data-mechanism fusion inversion model; S202: Constructing a state-space model includes a state transition model and an observation model. The physicochemical equations describing the evolution of the internal humidity field and chloride ion concentration field of concrete are used as the state transition model. The mathematical function of the state transition model is constructed to describe the dynamic evolution of the state vector from time k to time k. Based on the mathematical mapping relationship between state variables and the standardized bridge state feature vector set that can be actually obtained, an observation model of the connection state is established. N state vector samples are randomly generated to form the initial state set. ; S203: Based on the initial state set and dynamic data assimilation, the output of the running data-mechanism fusion inversion model is obtained. The output layer is the probabilistic estimate set of the output degradation driving state at the current time.

[0020] Dynamic data assimilation is performed based on the state-space model to obtain a probabilistic estimate set of the degradation-driving state at the current time, specifically including the following: For each individual in the current state set Based on the state transition model, the process proceeds to the next time step to obtain the predicted driving state. ; Each predicted driving state individual The corresponding predicted observation values ​​are calculated using the observation model. The observation vectors obtained after actual new collection and standardization Input, by comparing the actual observed values ​​with the set of all predicted observed values ​​{ The mean and covariance of} are used to calculate the optimal gain coefficient, and this coefficient is then used to correct the individual predicted driving state. , This is called reverse update, based on the observed residuals. Inverse optimization is used to estimate the internal state of an individual. The new set of states obtained after the update { , which is the probabilistic estimate set of the degenerate driving state obtained at time k, i={1,2,……,N}.

[0021] Each sample in the probabilistic estimate set of degradation-driving states is input into a predetermined bridge degradation mechanism model to perform long-term deterministic simulations and predict the future performance degradation trajectory of the bridge. The specific process is as follows: S400. Obtain the individual internal control states in the probabilistic estimation set of degradation-driven states, bind them with deterministic future environmental scenario assumptions, and form a complete future evolution input scenario package. Starting from the initial internal state sample, under the drive of the future environmental scenario, run the degradation mechanism model to perform long-term stepwise time accumulation and calculate the evolution of the internal state field at each time point. S401. At each preset output time point in the simulation process, based on the currently simulated internal state, the performance index values ​​calculated at each output time point are connected in chronological order to determine the future performance degradation path. S402. Aggregate the N simulated deterministic future performance degradation paths by time point and calculate the degradation index statistics for each time point; ;

[0022] for Expected value of performance metrics at any given time. For the nth path in Performance values ​​at any given time for The standard deviation of the time-space performance metric, where N is the number of simulated paths; State evolution formula:

[0023] This represents the current internal state vector, such as humidity and chloride ion concentration. For the time step of environmental simulation, Environment input vector, Let be the internal state vector at the next time step. The degradation mechanism function is a vector-valued function that is usually nonlinear;

[0024] M is a mapping function. In order to be in Real-time performance metrics, such as reliability and crack width;

[0025] In order to be in The p-quantile values ​​of the bridge degradation performance index at any given time, with .

[0026] Multiple future performance degradation paths are generated, and all paths are integrated to form a degradation trajectory cloud map that reflects the uncertainty of prediction. The specific process is as follows: A degradation trajectory cloud map is constructed with time as the horizontal axis and performance index as the vertical axis. The cloud map is defined by the region of quantile trajectory as the prediction interval, and the mean trajectory and median trajectory can be optionally superimposed as the central trend line to form a future performance degradation trajectory cloud map that intuitively reflects the uncertainty of prediction. Plot the lower bound percentile curve and the upper bound percentile curve, and fill the area between the two percentile curves to form the prediction interval band; Within the prediction interval, draw the mean curve or median curve as the central trend line.

[0027] Based on the probability distribution of future performance degradation trajectories in the degradation trajectory cloud map, a stochastic optimization model is constructed with conditionally triggered maintenance rules as decision variables and minimizing the expected lifecycle cost as the multi-objective. The optimal maintenance strategy set is then output by solving the model. The specific process is as follows: S600: Transform maintenance decisions into a dynamic condition-triggered mode, setting the maintenance trigger condition as the degradation index exceeding the warning value; This setting is based on the monitoring and assessment of equipment or structural performance degradation. When the performance degradation reaches a certain level and may threaten the normal operation and safety of the equipment or structure, maintenance action is triggered. The warning value is the average of historical maintenance time. S601. Establish mathematical programming conditions with the single objective of minimizing the expected total life cycle cost, calculate the expected total cost based on the probability distribution of the degradation trajectory, and output a specific set of optimal condition-triggered maintenance strategies defined by the optimal parameters.

[0028] Develop a detailed maintenance plan based on the optimal maintenance strategy set, clearly defining the maintenance time, maintenance measures, and maintenance scope; continuously adjust and optimize the maintenance strategies during the implementation of the maintenance plan, with specific steps as follows: Each strategy is transformed into a correspondence between monitoring and early warning values ​​and maintenance actions. Based on the early warning values ​​set in the optimal strategy set, when the real-time monitoring data exceeds the early warning value, the corresponding maintenance action is triggered. According to the triggering conditions, the specific scope of each measure, the area of ​​maintenance, and the degree of maintenance are planned in advance. The real-time monitoring and periodic testing data from step one are continuously received and analyzed, and compared in real time with the various trigger values ​​preset in the strategy set. An alarm is automatically generated to produce a maintenance work order. After maintenance is carried out, the latest and most comprehensive monitoring data are input into the data-mechanism fusion inversion model in step two at fixed intervals to re-estimate the internal degradation driving state of the bridge.

[0029] The size of the interval and threshold is set to facilitate comparison. The size of the threshold depends on the amount of sample data and the number of bases set by those skilled in the art for each set of sample data; as long as it does not affect the ratio between the parameter and the quantized value.

[0030] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation. In the two embodiments provided in this application, it should be understood that the disclosed methods can be implemented in other ways; for example, the device embodiments described above are merely illustrative, and the division of modules is merely a logical functional division. In actual implementation, there may be other division methods, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed; another point is that the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces, and the indirect coupling or communication connection of devices or modules may be electrical, mechanical or other forms. The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for preventive maintenance decision of concrete bridge based on deterioration prediction, characterized in that, Includes the following steps: Step 1: Collect degradation characteristic data of concrete bridges using bridge inspection equipment. The degradation characteristic data includes time-series data from automated monitoring, discrete data from periodic inspections, and historical environmental data. The degradation characteristic data is then cleaned, spatiotemporally aligned, and standardized to generate a standardized bridge state feature vector set. Step 2: Based on the standardized bridge state feature vector set, the degradation driving state inside the bridge is estimated through the data-mechanism fusion inversion model. The standardized bridge state feature vector set is used as input features, and the humidity field and chloride ion concentration field inside the concrete are used as state variables. Dynamic data assimilation is performed according to the state space model to obtain the probabilistic estimate set of the degradation driving state at the current moment. Step 3: Input each sample in the probabilistic estimation set of degradation driving states into the predetermined bridge degradation mechanism model, perform long-term deterministic simulation to predict the future performance degradation trajectory of the bridge, generate multiple future performance degradation paths, and integrate all paths to form a degradation trajectory cloud map that reflects the uncertainty of the prediction. Step 4: From the probability distribution of future performance degradation trajectories in the degradation trajectory cloud map, construct a stochastic optimization model with conditionally triggered maintenance rules as decision variables and minimizing expected lifecycle costs as a multi-objective, and output the optimal maintenance strategy set by solving the model. Step 5: Develop a detailed maintenance plan based on the optimal maintenance strategy set, specifying the maintenance time, maintenance measures and maintenance scope. During the implementation of the maintenance plan, continuously adjust and optimize the maintenance strategies dynamically.

2. The decision-making method for preventive maintenance of concrete bridges based on degradation prediction according to claim 1, characterized in that, The standardized bridge state feature vector set is generated through the following process: S100: Automatically acquire continuous time-series data from the deployed sensor network at a fixed frequency, and label each data point with a precise timestamp and sensor spatial location coordinates. Integrate the acquired discrete data, and associate each data point with the detection date, detection point number, and the component number to which it belongs. S102. Create a mapping table that associates sensor ID, detection point, component number with bridge, and perform data cleaning and continuous proxy sequence generation. From the cleaned continuous sequence, extract statistical and mechanistic features that can directly and sensitively reflect the process of concrete degradation, including concrete internal and surface temperature and humidity data, chloride ion concentration data, and bridge structure displacement and deformation data. For degradation characteristic data, not only are absolute values ​​recorded, but also the changes compared to the previous detection, the ratio of the historical maximum value, and the mean and standard deviation of the statistical distribution parameters of data from different detection points on the same component are calculated. S103. The standardized feature vectors of all components at the same time point are concatenated with the global environment feature vector to form a complete, standardized, spatiotemporally aligned bridge state feature vector set.

3. The decision-making method for preventive maintenance of concrete bridges based on degradation prediction according to claim 1, characterized in that, The degradation driving state inside concrete bridges is estimated by running a data-mechanism fusion inversion model. The specific process is as follows: S201: Discretize the pore relative humidity distribution H and free chloride ion concentration distribution C within the concrete cover into a state vector, denoted as X= , as input to the running data-mechanism fusion inversion model; S202: Constructing a state-space model includes a state transition model and an observation model. The physicochemical equations describing the evolution of the internal humidity field and chloride ion concentration field of concrete are used as the state transition model. The mathematical function of the state transition model is constructed to describe the dynamic evolution of the state vector from time k-1 to time k. Based on the mathematical mapping relationship between state variables and the standardized bridge state feature vector set that can be actually obtained, an observation model of the connection state is established. N state vector samples are randomly generated to form the initial state set. ; S203: Based on the initial state set and dynamic data assimilation, the output of the running data-mechanism fusion inversion model is obtained. The output layer is the probabilistic estimate set of the output degradation driving state at the current time.

4. The decision-making method for preventive maintenance of concrete bridges based on degradation prediction according to claim 3, characterized in that, Dynamic data assimilation is performed based on the state-space model to obtain a probabilistic estimate set of the degradation-driving state at the current time, specifically including the following: For each individual in the current state set Based on the state transition model, the process proceeds to the next time step to obtain the predicted driving state. ; Each predicted driving state individual The corresponding predicted observation values ​​are calculated using the observation model. The observation vectors obtained after actual new collection and standardization Input, by comparing the actual observed values ​​with the set of all predicted observed values ​​{ The mean and covariance of} are used to calculate the optimal gain coefficient, and this coefficient is then used to correct the individual predicted driving state. ， This is called reverse update, based on the observed residuals. Inverse optimization is used to estimate the internal state of an individual. The new set of states obtained after the update { , which is the probabilistic estimate set of the degenerate driving state obtained at time k, i={1,2,……,N}.

5. The decision-making method for preventive maintenance of concrete bridges based on degradation prediction according to claim 1, characterized in that, Each sample in the probabilistic estimate set of degradation-driving states is input into a predetermined bridge degradation mechanism model to perform long-term deterministic simulations and predict the future performance degradation trajectory of the bridge. The specific process is as follows: S400. Obtain the individual internal control states in the probabilistic estimation set of degradation-driven states, bind them with deterministic future environmental scenario assumptions, and form a complete future evolution input scenario package. Starting from the initial internal state sample, under the drive of the future environmental scenario, run the degradation mechanism model to perform long-term stepwise time accumulation and calculate the evolution of the internal state field at each time point. S401. At each preset output time point in the simulation process, based on the currently simulated internal state, the performance index values ​​calculated at each output time point are connected in chronological order to determine the future performance degradation path. S402. Aggregate the N simulated deterministic future performance degradation paths by time point and calculate the degradation index statistics for each time point; ； ； for Expected value of performance metrics at any given time. For the nth path in Performance values ​​at any given time for The standard deviation of the time-matter performance metric, where N is the number of simulated paths.

6. The decision-making method for preventive maintenance of concrete bridges based on degradation prediction according to claim 1, characterized in that, Multiple future performance degradation paths are generated, and all paths are integrated to form a degradation trajectory cloud map that reflects the uncertainty of prediction. The specific process is as follows: A degradation trajectory cloud map is constructed with time as the horizontal axis and performance index as the vertical axis. The cloud map is defined by the region of quantile trajectory as the prediction interval, and the mean trajectory and median trajectory can be optionally superimposed as the central trend line to form a future performance degradation trajectory cloud map that intuitively reflects the uncertainty of prediction. Plot the lower bound percentile curve and the upper bound percentile curve, and fill the area between the two percentile curves to form the prediction interval band; Within the prediction interval, draw the mean curve or median curve as the central trend line.

7. The decision-making method for preventive maintenance of concrete bridges based on degradation prediction according to claim 1, characterized in that, Based on the probability distribution of future performance degradation trajectories in the degradation trajectory cloud map, a stochastic optimization model is constructed with conditionally triggered maintenance rules as decision variables and minimizing the expected lifecycle cost as the multi-objective. The optimal maintenance strategy set is then output by solving the model. The specific process is as follows: S600: Transform maintenance decisions into a dynamic condition-triggered mode, setting the maintenance trigger condition as the degradation index exceeding the warning value; S601. Establish mathematical programming conditions with the single objective of minimizing the expected total life cycle cost, calculate the expected total cost based on the probability distribution of the degradation trajectory, and output a specific set of optimal condition-triggered maintenance strategies defined by the optimal parameters.

8. The decision-making method for preventive maintenance of concrete bridges based on degradation prediction according to claim 1, characterized in that, Develop a detailed maintenance plan based on the optimal maintenance strategy set, specifying the maintenance time, measures, and scope; continuously adjust and optimize the maintenance strategies during the implementation of the maintenance plan, with specific steps as follows: Each strategy is transformed into a correspondence between monitoring and early warning values ​​and maintenance actions. Based on the triggering conditions, the specific scope of each measure is planned in advance. Real-time monitoring and periodic testing data from step one are continuously received and analyzed. The data are compared in real time with the various triggering values ​​preset in the strategy set, and an alarm is automatically generated to produce a maintenance work order. After maintenance is carried out, the latest and most comprehensive monitoring data are input into the data-mechanism fusion inversion model in step two at fixed intervals to re-estimate the internal degradation driving state of the bridge.

Images