Infrastructure deterioration simulation and maintenance optimization decision method, device and equipment

Abstract

The application provides a kind of infrastructure deterioration simulation and maintenance optimization decision method, device and equipment, it is related to infrastructure operation and maintenance technical field, wherein, the method comprises: obtaining the multi-source data related to the component of infrastructure;With the component in building information model, the space-time correlation of the multi-source data is carried out, and the data fusion graph of component level is constructed;Based on the data fusion graph, the future deterioration state of the component is predicted using the pre-constructed physical-data double-driven deterioration state prediction model, and the component health degree is obtained;Based on the component health degree, a multi-objective maintenance decision optimization method based on stochastic simulation is used to generate a set of maintenance schemes representing the performance of different maintenance strategies. Through the present application, data silos can be broken, prediction accuracy can be improved, and scientific and quantitative basis for maintenance decision can be provided through multi-objective optimization method, which helps to improve operation and maintenance efficiency.

Description

Technical Field

[0001] This invention relates to the field of infrastructure operation and maintenance technology, and in particular to a method, apparatus and equipment for infrastructure deterioration simulation and maintenance optimization decision-making. Background Technology

[0002] Rail transit infrastructure operates under complex loads and harsh environments for extended periods. Key components such as rails, sleepers, and bridge bearings are prone to fatigue, cracks, deformation, and other deterioration damage, directly threatening operational safety. Existing operation and maintenance technologies face three major bottlenecks: (1) Disconnection between multi-source heterogeneous data and structural ontology: Monitoring data, inspection records, maintenance history and other information are stored in a scattered manner, lacking accurate spatiotemporal correlation with three-dimensional components, making it impossible to construct a component-level full life cycle data profile, and making it difficult to deeply explore the value of the data; (2) Insufficient accuracy of degradation prediction model: Pure data-driven model has no physical mechanism constraints, and the prediction is unstable when the data is sparse or the operating conditions change abruptly; Pure physical model parameters rely on idealized assumptions and cannot adapt to the dynamic degradation law of the real service environment, and the prediction results deviate significantly from the actual measurements. (3) Lack of quantitative support for maintenance decisions: Maintenance plans rely on experience and fixed procedures, and do not comprehensively consider multiple objectives such as cost, safety risks, and resource constraints. They cannot achieve forward-looking simulation and comparison, and it is difficult to achieve the optimal operation and maintenance benefits throughout the entire life cycle.

[0003] Therefore, existing technologies suffer from drawbacks such as the disconnect between multi-source data and structural ontology, insufficient accuracy of degradation prediction models, and lack of quantitative support for maintenance decisions. Summary of the Invention

[0004] This invention provides a method, apparatus, and equipment for infrastructure degradation simulation and maintenance optimization decision-making, which addresses the shortcomings of existing technologies such as the disconnect between multi-source data and structural ontology, insufficient accuracy of degradation prediction models, and lack of quantitative support for maintenance decisions.

[0005] In a first aspect, the present invention provides an infrastructure degradation simulation and maintenance optimization decision-making method, comprising: Acquire multi-source data related to the components of the infrastructure; The data will be spatiotemporally correlated with the components in the building information model to construct a component-level data fusion map. Based on the data fusion map, a pre-built physical-data dual-driven degradation state prediction model is used to predict the future degradation state of the component and obtain the component health. Based on the health status of the components, a multi-objective maintenance decision optimization method based on stochastic simulation is used to generate a set of maintenance schemes that characterize the performance of different maintenance strategies.

[0006] According to the infrastructure degradation simulation and maintenance optimization decision-making method provided by the present invention, the multi-source data includes real-time monitoring data, historical inspection data, and maintenance history data.

[0007] According to the present invention, an infrastructure deterioration simulation and maintenance optimization decision-making method is used to spatiotemporally correlate the multi-source data with components in the building information model to construct a component-level data fusion map, including: The building information model is then subjected to lightweight processing and semantic enhancement. Assign a unique identifier to the component, expand the semantic attribute slots of the building information model, and explicitly define the data carrying dimensions; The multi-source data is anchored to the component using timestamps and spatial coordinates to generate the data fusion map.

[0008] According to the infrastructure degradation simulation and maintenance optimization decision-making method provided by the present invention, a degradation state prediction model driven by both physical and data aspects is constructed, including: A physical mechanism model is constructed using damage mechanics as the core framework. The physical mechanism model is subjected to dynamic data calibration and dynamic state calibration to obtain a physical-data dual-driven degradation state prediction model.

[0009] According to the infrastructure degradation simulation and maintenance optimization decision-making method provided by the present invention, the dynamic data calibration is achieved by constructing a dynamic parameter correction function; The parameter dynamic correction function uses the real-time monitoring data from the multi-source data to dynamically adjust the model parameters in the physical mechanism model.

[0010] According to the infrastructure degradation simulation and maintenance optimization decision-making method provided by the present invention, the dynamic state calibration is achieved by introducing an unscented Kalman filter algorithm; The unscented Kalman filter algorithm includes: Sigma point sampling is used to extract sample points that can accurately represent the probability distribution of the model state of the physical mechanism model; The sample points are weighted and summed to integrate the scattered sample point prediction values ​​into a single state prediction value. The predicted state is corrected by dynamically adjusting the correction weights using Kalman gain.

[0011] According to the present invention, an infrastructure degradation simulation and maintenance optimization decision-making method is provided. Based on the component health, a multi-objective maintenance decision optimization method based on stochastic simulation is used to generate a set of maintenance schemes characterizing the performance of different maintenance strategies, including: The Monte Carlo simulation method is used to define the maintenance strategy space and to simulate and deduce multiple preset maintenance strategies. Construct a multi-objective optimization function with both lifecycle cost and safety risk as optimization objectives; The maintenance scheme set is determined by a multi-objective optimization algorithm; wherein the maintenance scheme set is a Pareto optimal scheme set.

[0012] According to the present invention, an infrastructure degradation simulation and maintenance optimization decision-making method is provided, the method further includes: feeding back real-time monitoring data after maintenance to the data fusion map and updating the degradation state prediction model.

[0013] Secondly, the present invention also provides an infrastructure degradation simulation and maintenance optimization decision-making device, comprising: The acquisition module is used to acquire multi-source data related to the components of the infrastructure; The association module is used to spatiotemporally associate the multi-source data with the components in the building information model to construct a component-level data fusion map; The prediction module is used to predict the future degradation state of the component based on the data fusion map and using a pre-built physical-data dual-driven degradation state prediction model, so as to obtain the component health. The decision module is used to generate a set of maintenance schemes that characterize the performance of different maintenance strategies based on the health status of the components and a multi-objective maintenance decision optimization method based on stochastic simulation.

[0014] Thirdly, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the infrastructure degradation simulation and maintenance optimization decision-making method as described in the first aspect above.

[0015] Fourthly, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the infrastructure degradation simulation and maintenance optimization decision-making method as described in the first aspect above.

[0016] Fifthly, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the infrastructure degradation simulation and maintenance optimization decision-making method as described in the first aspect above.

[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. Improved data fusion accuracy: Achieves precise anchoring of multi-source data and BIM components, constructs a complete component-level full lifecycle data profile, breaks down data silos, and improves data utilization by more than 60%, providing comprehensive and accurate data support for degradation prediction and decision optimization; 2. More accurate degradation prediction: The physical-data dual-drive model combines physical mechanism constraints with dynamic data calibration, overcoming the shortcomings of traditional single models. The prediction error is reduced to less than 5%, which can adapt to complex working conditions and data sparse scenarios, and improves the reliability of degradation state prediction. 3. Quantitative and scientific decision-making: Through Monte Carlo simulation and Pareto optimization, the maintenance strategy is quantitatively compared and selected, and the cost-risk multi-objective trade-off is achieved. Compared with traditional experience-based decision-making, the operation and maintenance cost is reduced by 15%-25%, the safety risk is reduced by 30%, and the operation and maintenance benefits throughout the entire life cycle are optimized. 4. Upgraded Operation and Maintenance Mode: The digital twin closed-loop decision-making system connects the entire chain of perception, modeling, simulation, decision-making, and feedback, realizing an intelligent transformation from "passive emergency repair" to "proactive prediction, scientific decision-making, and continuous optimization." It adapts to the operation and maintenance needs of rail transit infrastructure throughout its entire life cycle, improving operation and maintenance efficiency and safety. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0019] Figure 1 This is a flowchart of the infrastructure degradation simulation and maintenance optimization decision-making method provided by the present invention; Figure 2 This is a schematic diagram illustrating the specific process of making maintenance optimization decisions in an embodiment of the present invention; Figure 3 This is a structural block diagram of the infrastructure degradation simulation and maintenance optimization decision-making device provided by the present invention; Figure 4 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0021] Please see Figure 1 This application provides a method for infrastructure degradation simulation and maintenance optimization decision-making, aiming to solve problems in existing technologies such as the disconnect between infrastructure operation and maintenance data and the structural entity, insufficient accuracy of degradation prediction models, and lack of quantitative support for maintenance decisions that rely on experience. This method focuses on key components (such as rails, bridge bearings, and tunnel linings) in infrastructure such as rail transit, highways, and municipal works. By constructing a digital twin closed-loop decision-making system, it achieves dynamic optimization of the entire process from data acquisition to intelligent operation and maintenance.

[0022] Figure 1 This is a flowchart of the infrastructure degradation simulation and maintenance optimization decision-making method provided by the present invention, such as... Figure 1 As shown, the method includes: Step S101: Obtain multi-source data related to the components of the infrastructure; Step S102 involves spatiotemporally associating the multi-source data with the components in the Building Information Modeling (BIM) system to construct a component-level data fusion map. Step S103: Based on the data fusion map, a pre-built physical-data dual-driven degradation state prediction model is used to predict the future degradation state of the component and obtain the component health. Step S104: Based on the component health, a multi-objective maintenance decision optimization method based on stochastic simulation is used to generate a set of maintenance schemes that characterize the performance of different maintenance strategies.

[0023] Figure 2 This is a schematic diagram illustrating the specific process of making maintenance optimization decisions in an embodiment of the present invention, such as... Figure 2 As shown, the method in this embodiment first performs a step of acquiring multi-source data related to the components of the infrastructure. This step forms the data foundation for the entire decision-making process, corresponding to... Figure 2 The data acquisition phase is designed to break the data silo dilemma caused by the scattered storage and inconsistent formats of various types of data in traditional operations and maintenance. By proactively and systematically collecting all information related to the performance status of components, it provides comprehensive and original input for the subsequent establishment of a full lifecycle data profile of components, solving the fundamental problem of being unable to conduct in-depth analysis due to incomplete or uncorrelated data.

[0024] Next, this method spatiotemporally correlates multi-source data with components in the Building Information Model (BIM) to construct a component-level data fusion map. This step corresponds to... Figure 2The construction of a component-level BIM data fusion map is a key step in the process. This step aims to address the core pain point of the disconnect between multi-source heterogeneous data and the 3D structural entity itself. Traditionally, monitoring data, inspection records, and other information lack precise and automated correlation with their corresponding specific component locations. This step strongly binds abstract data to concrete BIM components in both time and space, thereby constructing a unified data foundation centered on the components. This allows any data to be traced back to its constituent component and the time it occurred, enabling in-depth mining of data value.

[0025] Subsequently, based on the data fusion map, a pre-built physical-data dual-driven degradation state prediction model is used to predict the future degradation state of the component, thus obtaining the component's health. This step corresponds to... Figure 2 This paper introduces a physical-data-driven dual-engine degradation prediction model and quantifies component health. This step aims to address the shortcomings of traditional degradation prediction models, such as insufficient accuracy and poor generalization ability. By integrating the universality of physical mechanisms and the adaptability of data-driven approaches, a high-confidence prediction tool is created. This model not only describes the degradation patterns of components but also dynamically adjusts based on real-time data, ultimately outputting an intuitive and quantifiable component health index. This transforms complex degradation states into clear criteria for decision-making, solving the problem of ambiguous prediction results that are difficult to guide actual operation and maintenance.

[0026] Finally, based on component health, this method employs a multi-objective maintenance decision optimization method based on stochastic simulation to generate a set of maintenance schemes characterizing the performance of different maintenance strategies. This step corresponds to... Figure 2 The process involves multi-objective maintenance strategy optimization and the output of a Pareto optimal solution set. This step aims to address the problem that traditional maintenance decision-making relies on experience and lacks scientific quantitative comparison criteria. It uses the predicted future health level as input, simulates the potential long-term impacts of different maintenance interventions, and simultaneously considers the two core objectives of cost and safety risk. In this way, a set of optimal solutions, each with its own emphasis but no absolute superiority or inferiority, can be selected from numerous possibilities. This provides decision-makers with a clear decision space containing multiple trade-offs, thus achieving a shift from experience-based decision-making to scientific decision-making.

[0027] In a preferred embodiment, the multi-source data specifically includes real-time monitoring data, historical inspection data, and maintenance history data. This detailed definition establishes three main dimensions for data acquisition, corresponding to the component's real-time response, historical damage evolution, and history of human intervention. Real-time monitoring data (such as strain, acceleration, displacement, and temperature) reflects the component's immediate state under the current load; historical inspection data (such as crack and deformation records) constitutes the component's "medical record," documenting its long-term deterioration process; and maintenance history data records the "treatment" measures taken for the component. By clearly defining these three types of data, it ensures that the collected information can fully cover the entire lifecycle of the component, from health to damage and then to repair. This provides complete and complementary data support for building accurate deterioration models and assessing maintenance costs, solving the problem of previous data acquisition methods being singular and unable to comprehensively depict the component's state.

[0028] For example, suppose the set of critical infrastructure components is:

[0029] in, C Represents a set of key components. Indicates the first i Each component N This formula represents the total number of components. It clarifies that the research object of this method is a single component and a set of components, defining the scope for subsequent component-level data fusion, degradation modeling, and maintenance decision-making. This enables refined "component-level" management, solves the problem of insufficient precision in traditional "holistic" management, and lays the foundation for accurate correlation between multi-source data and components.

[0030] Multi-source data for any component It contains three types of data: 1. Real-time monitoring data:

[0031] in, Indicates the first i Each component t Real-time monitoring data at all times t Indicates time, Indicates the first k Sensor-like sensors t Monitoring values ​​at any given time K This indicates the total number of sensor types.

[0032] This formula realizes the structured definition of real-time monitoring data, accurately correlates scattered sensor data with components and time, effectively solves the problem of real-time monitoring data being disconnected from components, and provides real-time and accurate data input for the subsequent dynamic calibration of parameters of the dual-drive model, ensuring the timeliness of model prediction.

[0033] 2. Historical inspection data:

[0034] in, Indicates the first i Historical inspection data for each component Indicates the inspection time. Indicates the type of disease. Indicates the severity of the disease.

[0035] This method establishes a structured association between historical inspection data and components, fully recording the historical trajectory of component deterioration. It provides detailed historical data support for model parameter inversion and deterioration pattern analysis, solving the industry pain point of scattered and untraceable inspection data, and helping to accurately identify component deterioration trends.

[0036] 3. Maintenance history data:

[0037] in, This indicates maintenance history data. Indicates the repair time. Indicates the repair method. This indicates the cost of maintenance.

[0038] This method precisely binds maintenance data to components, forming a full lifecycle maintenance record for components. It provides quantitative data support for subsequent maintenance cost calculation and maintenance strategy optimization, solving the problem of maintenance data being disconnected from components and unable to provide a reliable basis for decision-making. It lays a data foundation for multi-objective optimization decision-making.

[0039] In another preferred embodiment, the specific process of spatiotemporally associating multi-source data with components in a Building Information Model (BIM) to construct a component-level data fusion graph includes: first, lightweighting and semantically enhancing the BIM; then, assigning unique identifiers to components, expanding the semantic attribute slots of the BIM, and explicitly defining the data carrying dimensions; finally, anchoring multi-source data with components using timestamps and spatial coordinates to generate a data fusion graph. This series of steps provides a systematic operational path for solving the problem of deep integration of data and BIM. Lightweighting the BIM solves the problem of the original model being too large and unsuitable for online applications and computation. Assigning unique identifiers provides each component with a unique "identity card," which is the foundation for accurate data linking. Expanding semantic attribute slots reserves "slots" within the BIM for storing and associating various types of data. Finally, through spatiotemporal anchoring, external monitoring, inspection, and maintenance data are precisely "inserted" into the corresponding component's "slot," ultimately forming a structured and semantic data fusion graph. This complete process thoroughly solves the problem of loose, nominal connections between data and components, achieving true integrated management of component-level data.

[0040] For example, the original building information model is geometrically simplified while retaining key structural features to generate a lightweight model with LOD300 accuracy. This ensures the model's engineering usability while reducing the complexity of subsequent data processing and model calculations. A unique identifier is assigned to each component. Expand semantic attribute slots and define data carrying dimensions (such as storage formats and association rules for real-time monitoring data, historical inspection data, and maintenance history data) to achieve a foundation for accurate association between BIM components and multi-source data.

[0041] By timestamp t Spatial coordinates Data and components C i Precise anchoring to construct a graph structure fusion graph. Nodes are components and their associated data, while edges... This diagram illustrates the spatial adjacency and mechanical transmission relationships between components. This graphical structure allows the health status of any component to be directly linked to its associated data, enabling integrated management of multi-source data. It provides a unified input for subsequent degradation prediction, completely resolving the technical pain point of the disconnect between multi-source data and the structural entity itself.

[0042] Furthermore, in a preferred embodiment, the process of constructing a physics-data dual-driven degradation state prediction model includes: firstly, constructing a physical mechanism model with damage mechanics as the core framework; and then performing data dynamic calibration and state dynamic calibration on the physical mechanism model. This structured construction method aims to address the inherent limitations of a single model (purely physical or purely data-driven). Firstly, constructing a physical mechanism model ensures that the prediction has a solid physical foundation and good interpretability, avoiding the problem of pure data models becoming "black boxes" and becoming unstable when data is sparse. The subsequent dual dynamic calibration, namely data dynamic calibration and state dynamic calibration, uses measured data to bridge the gap between the idealized assumptions of the physical model and the differences in the real environment. By continuously "correcting" the model, it can adapt to changes in the real service environment, thereby combining the rigor of the physical model with the flexibility of data-driven approaches, significantly improving prediction accuracy and reliability.

[0043] Taking the fatigue crack propagation of rails as an example, the Paris formula is used as the basic physical model, and the specific formula is as follows:

[0044] in, a This indicates the crack depth, in mm. N Indicates the number of load cycles (dimensionless). This represents the stress intensity factor amplitude, in MPa·m. 1 / 2 , C , m This indicates the material degradation parameter.

[0045] As a classic physical model describing fatigue crack propagation rate, the core function of this formula is to establish a power-law relationship between crack propagation rate and stress intensity factor amplitude, providing a solid physical mechanism constraint for degradation prediction, effectively solving the problem of uninterpretable pure data-driven models, and ensuring that model predictions have clear engineering physical significance.

[0046] In a more specific implementation, dynamic data calibration is achieved by constructing a dynamic parameter correction function. This function utilizes real-time monitoring data from multiple sources to dynamically adjust the model parameters in the physical mechanism model. This aims to address the problem that key coefficients such as material parameters in physical models are often treated as fixed constants, failing to reflect the effects of temperature and load changes in the real environment. By establishing a functional relationship between parameters and real-time monitoring data (such as temperature and stress), the model parameters are no longer static values ​​but dynamic quantities that change with the environment. For example, when the temperature rises, the fatigue resistance of the material may decrease, and the dynamic parameter correction function can quantify and reflect this change. In this way, real-time operating condition information is integrated into the core of the physical model, enabling the model to "sense" environmental changes in real time and make adjustments, greatly improving the model's adaptability to real service conditions.

[0047] For example, material degradation parameters in rail fatigue cracks. C , m Based on the characteristics affected by environment and load, a dynamic correction function is constructed:

[0048] in, These are the material reference parameters, obtained from laboratory tests, and serve as the basis for parameter correction. The component temperature (unit: °C) is obtained in real time through a temperature sensor; The longitudinal force between the wheel and rail (unit: kN) is acquired in real time through the wheel-rail force monitoring system. , The influence coefficients to be identified are obtained through joint inversion of historical monitoring data and crack detection data, and are used to quantify the effects of temperature and wheel-rail forces. The degree of impact. The core of this function is to dynamically correct material degradation parameters using real-time monitoring data (temperature, wheel-rail force). C Solving problems in traditional physical models C The problem of treating it as a constant and failing to adapt to real service environments is that its expression reflects the effects of increased temperature or wheel-rail forces. The exponential growth and accelerated crack propagation pattern closely matches the degradation characteristics of components in actual service, making the physical model more consistent with actual engineering scenarios.

[0049] in, These are the material reference parameters, obtained through laboratory testing. Stress amplitude fluctuation (unit: MPa) reflects the magnitude of load change and is acquired in real time by strain sensors; The influence coefficient to be identified is obtained through joint inversion of historical monitoring data and crack detection data, and is used to quantify the impact of stress amplitude fluctuations on... mThe degree of influence. This function corrects the material degradation parameters by real-time monitoring of stress amplitude fluctuations. m This further enhances the environmental adaptability of the physical model, and its linear relationship reflects the effect of increasing stress amplitude fluctuations. Consequently, the crack propagation rate increases. The actual pattern of enhanced sensitivity is analyzed by substituting real-time monitoring data into the function and comparing it with the above. The correction function enables dynamic adjustment of model parameters, making the prediction results more consistent with real service conditions and effectively solving the problem of fixed parameters and large prediction deviations in traditional physical models.

[0050] In another, more specific implementation, to further utilize periodically measured data (such as crack depth obtained from ultrasonic testing) to correct the model state (crack depth, material parameters), an Unscented Kalman Filter (UKF) algorithm is introduced. This algorithm includes: extracting sample points that accurately represent the probability distribution of the model state through Sigma point sampling; weighting and summing the sample points to integrate the scattered predicted values ​​into a single state prediction; and dynamically adjusting the correction weights through Kalman gain to correct the predicted state. The purpose of introducing the Unscented Kalman Filter is to address the discrepancy between the physical model's predicted values ​​and periodically measured values ​​(such as crack depth obtained from manual testing). The Unscented Kalman Filter uses a "prediction-update" mechanism, using measured data as the basis for correction, to continuously calibrate the model's state prediction. Its advantage over traditional Kalman filters lies in handling nonlinear problems through Sigma point sampling, avoiding errors caused by linearization, and enabling more accurate fusion of measured data. Through this mechanism, the model achieves self-correction and learning capabilities, ensuring the accuracy and convergence of long-term predictions and solving the problem that model predictions gradually deviate from reality over time.

[0051] For example, define a system state vector. subscript Indicates the first At each detection point, the crack depth and two material degradation parameters are included, enabling comprehensive monitoring and calibration of the model's core state. UKF achieves state updates through three steps: Sigma point sampling, prediction, and updating. The core formulas and their functions are as follows: The specific formula for Sigma point sampling is as follows:

[0052] in, Indicates the sample points extracted. This is the state estimate from the previous time step. The state covariance at the previous time step. State dimension (corresponding to crack depth) ,parameter ,parameter ), For Sigma point sampling coefficients, This represents the sampling point number. The core function of this formula is to extract from the state estimate and its covariance at the previous time step. These sample points accurately represent the probability distribution of the state, effectively solving the problem of large linearization error in traditional Kalman filtering in nonlinear systems, and laying a solid foundation for accurate prediction of the subsequent state.

[0053] The core formula for the prediction step is as follows:

[0054] in, This represents the predicted state value at the current moment. The predicted sample points derived from the forward extrapolation for each Sigma point. The weights of the sample points are used. This formula integrates the scattered predicted values ​​of all predicted sample points into a single accurate state prediction by weighted summation of all predicted sample points, effectively improving the reliability of the prediction results. This is further enhanced by combining the state transition function. By extrapolating the Sigma point forward using a physical mechanism model and a dynamic parameter correction function, accurate forward prediction of the state can be achieved.

[0055] The core formula for the update step is as follows:

[0056] in, For Kalman gain, For the first The measured crack depth obtained from this test. For the observed predicted value, This is the updated state estimate. The core of this formula is to use the deviation between the measured detection data and the observed prediction value, and dynamically adjust the correction weights through Kalman gain to accurately correct the predicted state, realizing a closed loop of "prediction-measurement-correction". This effectively integrates the measured data into the model, significantly reduces the prediction error, and solves the problem of large deviation between traditional models and measured data. At the same time, the reliability of the state estimate is ensured by the covariance update formula, ensuring the continuous and stable operation of the UKF algorithm.

[0057] Through the three core steps of UKF described above, each new detection data will correct the model's state (crack depth) and parameters. , This ensures that the model always fits the actual service conditions, greatly improving the accuracy and reliability of degradation prediction.

[0058] Furthermore, in order to intuitively reflect the health status of components and provide a quantitative basis for subsequent maintenance decisions, a health index for the component at future moments is defined, transforming the abstract crack depth into a quantitative indicator that can be directly used for decision-making:

[0059] in, The predicted future crack depth (unit: mm) is obtained by extrapolation using the UKF algorithm, reflecting the future deterioration state of the component; The critical crack depth (unit: mm) allowed for a component is given by the design specifications and is the critical value at which the component reaches a dangerous state. For the health of components, For the current moment, The formula transforms the predicted crack depth into a health index within the [0,1] interval, quantifying the component's health status and effectively addressing the problem that traditional qualitative health assessments cannot provide precise data for maintenance decisions. The health index is used to predict the time interval. This indicates that the component is completely healthy. This indicates that the component has reached critical damage and requires immediate repair. This indicator can intuitively reflect the remaining lifespan of the component, providing a clear quantitative basis for the formulation of subsequent maintenance strategies (such as determining maintenance thresholds).

[0060] In a preferred embodiment, the specific process of generating a maintenance scheme set using a multi-objective maintenance decision optimization method based on stochastic simulation includes: defining a maintenance strategy space using Monte Carlo simulation and performing simulation deductions on multiple preset maintenance strategies; constructing a multi-objective optimization function with life-cycle cost and safety risk as dual optimization objectives; and determining the maintenance scheme set through a multi-objective optimization algorithm, wherein the maintenance scheme set is a Pareto optimal scheme set. This process aims to elevate maintenance decision-making from qualitative experience-based judgment to quantitative scientific optimization. First, through Monte Carlo simulation, considering uncertainties such as future loads and environment, thousands to tens of thousands of simulations are performed on each maintenance strategy (such as "polishing when health is below 0.7" or "replacing when health is below 0.5") to obtain the statistical expected values ​​of its cost and risk, solving the problem that single-path deduction cannot reflect randomness. Second, constructing a dual-objective optimization function that includes cost and risk ensures the comprehensiveness of the decision and avoids the one-sidedness of simply pursuing low cost while ignoring safety, or prioritizing safety regardless of cost. Finally, by using multi-objective optimization algorithms (such as NSGA-II) to select a set of Pareto optimal solutions, decision-makers are provided with a series of mutually exclusive "dilemma" optimal choices, which helps to improve operational efficiency.

[0061] For example, define a maintenance strategy space. ,in Total number of strategies. Individual strategy. Includes: preventative maintenance threshold (Repair is triggered when the component's health level falls below this value), repair method (e.g., grinding, replacement, reinforcement), and resource input. Different maintenance strategies correspond to different maintenance timings and methods, covering common maintenance scenarios in engineering, and providing a comprehensive strategy sample for subsequent simulation and comparison.

[0062] With life-cycle cost and safety risk as dual optimization objectives, a multi-objective optimization function is constructed to achieve a balance between the two, overcoming the limitations of traditional maintenance decision-making with a single objective (such as considering only cost). The specific formula is as follows:

[0063] in, This indicates that the strategy to be optimized is selected from all maintenance strategies. For strategy The corresponding total lifecycle cost, For strategy The corresponding cumulative security risks, This is a dual-objective optimization function vector. The function clarifies the optimization objective and scope, with the core objective being to achieve a multi-objective trade-off in maintenance strategies. This minimizes both lifecycle costs and safety risks, breaking the limitations of traditional "single-objective" maintenance decisions and maximizing operational efficiency. It provides core guidance for subsequent strategy simulation and optimization.

[0064] Objective 1: Total lifecycle cost, quantifying the economics of maintenance strategies:

[0065] in, The cost of repair (unit: yuan) depends on the repair method and the amount of resources invested. For example, the cost of replacing components is higher than the cost of grinding. Losses due to operational interruptions (unit: yuan) are incurred due to maintenance taking up maintenance windows or causing operational delays. For example, losses due to maintenance at night are lower than those during the day. The discount rate (dimensionless) discounts future costs to present value, which aligns with conventional methods of engineering economic analysis. The planning period (unit: years) is determined based on the service life of the infrastructure. This is the expectation operator. The formula comprehensively reflects the economics of the strategy by calculating the discounted expected value of all maintenance-related costs within the planning period, avoiding the neglect of long-term losses due to only considering short-term costs, and providing accurate quantitative basis for the economic evaluation of the strategy.

[0066] Objective 2: Accumulated safety risks and quantifying the safety of maintenance strategies:

[0067] in, The component health is given by a health quantification formula, reflecting the real-time health status of the component. The failure loss function (unit: yuan / year) represents the economic loss that may be caused per unit time when the health level is lower than the safe value (including accident risk, additional maintenance, etc.). The lower the health level, the greater the failure loss. It is an indicator function, which takes the value 1 when the condition is met and 0 otherwise. That is, the safety risk loss is calculated only when the health level is lower than the safety threshold. The health threshold is used to determine whether a component is in a high-risk state and requires close monitoring when its health level is below this value. For the expectation operator, The planning period is defined by this formula. By calculating the time-weighted expected loss of components in a high-risk state during the planning period, this formula comprehensively reflects the safety risks of the strategy, avoids safety hazards caused by excessive pursuit of cost savings, and provides accurate quantitative basis for the safety assessment of the strategy.

[0068] Due to the randomness of future loads, environment, and testing errors, the degradation path of components is...

[0069] Due to inherent uncertainties, a single simulation cannot accurately evaluate the actual effectiveness of a strategy. Therefore, simulations are used to model stochastic scenarios. The core formulas and their functions are as follows:

[0070]

[0071] in, For the first One maintenance strategy, The number of simulations (usually) ), For the first The total cost of the simulation For the first The total risk of this simulation , respectively strategy The expected lifecycle cost and expected cumulative safety risk are calculated. This set of formulas, using the sample mean from multiple random simulations, effectively reduces the impact of random factors on cost and risk assessment, yielding more reliable target expectations, eliminating random interference, ensuring the accuracy of cost and risk assessment, and providing a reliable basis for subsequent strategy comparisons.

[0072] The core statement of the Pareto dominance criterion is as follows: Strategy Pareto Domination Strategy If and only if: and And at least one of them is strictly true.

[0073] in, , For any two different maintenance strategies, , These represent the expected cost and expected risk of a strategy, respectively. The core function of this criterion is to screen out "non-dominated strategies"—that is, strategies that do not have any other capability to reduce one objective (cost or risk) without increasing another. All strategies not dominated by any other strategy constitute the Pareto front. The system outputs the set of optimal solutions on this front, from which decision-makers can choose the final maintenance solution based on their own preferences (such as prioritizing cost savings or risk aversion). This solves the problem of traditional maintenance decision-making lacking quantitative comparison and selection, and achieves scientific decision-making.

[0074] In an optional implementation, the method further includes feeding back real-time monitoring data after maintenance to the data fusion map to update the degradation state prediction model. This step corresponds to... Figure 2 The system establishes a closed-loop path for maintenance effect feedback and data updates. Its core objective is to build a dynamic decision-making system capable of self-learning and continuous evolution. When a maintenance plan is implemented, the state of a component changes, and its subsequent degradation behavior may deviate from predictions. By feeding back new monitoring data after maintenance into the data fusion map, parameters in the physical-data dual-driven degradation state prediction model can be recalibrated, or the data can be used as new samples to update the state of the UKF (Underlying Key Functions). Through this closed-loop mechanism of data feedback and model updates, the system can continuously learn from actual operation and maintenance activities, continuously improving the accuracy of predictions and the effectiveness of decisions. This achieves a shift from one-time decision-making to dynamic continuous optimization, truly implementing the core concept of digital twins.

[0075] In this embodiment, a closed-loop process of "data acquisition → model update → strategy simulation → scheme execution → effect feedback" is constructed, which maps the BIM model, degradation prediction results, simulation data, and optimization schemes to the twin space in a unified manner, realizing intelligent and dynamic optimization of the entire process of infrastructure operation and maintenance, and promoting the transformation of operation and maintenance model.

[0076] Based on this, a digital twin dynamic decision-making closed-loop system has been formed, and the core functions of the system include: Real-time visualization of component-level health status: By linking the BIM model with real-time monitoring data and health prediction results, a three-dimensional visualization of the component's health status can be achieved, intuitively presenting information such as crack propagation and health status changes, making it easy for maintenance personnel to quickly grasp the component's status. Interactive comparison and quantitative analysis of multiple maintenance strategies: Display simulation results (cost, risk, degradation trajectory) of different maintenance strategies in twin space, support interactive comparison, and provide decision-makers with intuitive quantitative reference; Pareto Solution Optimization and Automatic Maintenance Scheduling: Based on the Pareto frontier solution set and combined with resource constraints (personnel, equipment, time), this feature automatically generates reasonable maintenance scheduling plans to improve operational efficiency. Repair effect feedback and iteration: The measured data after repair (such as crack depth and health status) is fed back to the data fusion map to update the degradation model parameters and simulation data, continuously optimize the model and repair strategy, and achieve closed-loop self-improvement.

[0077] In summary, compared with existing technologies, this method has significant technical advantages and innovations, the core innovations of which are as follows: 1. Innovative BIM Data Fusion Architecture: Using lightweight BIM as the spatiotemporal foundation, a component-level data fusion map is constructed. Through time and space coordinates, the precise anchoring of multi-source data and components is achieved, completely solving the industry pain point of the disconnect between multi-source heterogeneous data and the structural body, and realizing integrated data management and value mining. 2. Breakthrough in Dual-Drive Degradation Model: An innovative dual-drive degradation model of "physical mechanism constraint + dynamic data calibration" is proposed. Through the dynamic parameter correction function and UKF calibration, the physical interpretability and prediction accuracy of the model are taken into account. This overcomes the shortcomings of traditional pure data-driven models that are not interpretable and pure physical models that have large biases, thus improving the credibility and applicability of degradation prediction. 3. Multi-objective optimization decision innovation: Combining Monte Carlo simulation and Pareto optimization, it realizes quantitative simulation of maintenance strategies and multi-objective (cost, risk) trade-offs, and generates a Pareto optimal solution set, which solves the problem that traditional maintenance decision-making relies on experience and lacks quantitative support, and provides a reliable method for scientific operation and maintenance decision-making. 4. Innovation of Digital Twin Closed-Loop System: Construct a digital twin closed loop covering the entire process of "data acquisition - model update - strategy simulation - solution execution - effect feedback", connect the entire operation and maintenance chain, realize system self-optimization, promote the transformation of operation and maintenance mode from "passive emergency repair" to "proactive intelligent operation and maintenance", and adapt to the operation and maintenance needs of the whole life cycle.

[0078] The present invention also provides an infrastructure deterioration simulation and maintenance optimization decision-making device. The infrastructure deterioration simulation and maintenance optimization decision-making device provided by the present invention will be described below. The infrastructure deterioration simulation and maintenance optimization decision-making device described below can be referred to in correspondence with the infrastructure deterioration simulation and maintenance optimization decision-making method described above. Figure 3 This is a structural block diagram of the infrastructure degradation simulation and maintenance optimization decision-making device provided by the present invention, as shown in the figure. Figure 3 As shown, the device includes: Acquisition module 301 is used to acquire multi-source data related to the components of the infrastructure; The association module 302 is used to spatiotemporally associate multi-source data with components in the building information model to construct a component-level data fusion map; The prediction module 303 is used to predict the future deterioration state of a component based on a data fusion map and a pre-built physical-data dual-driven deterioration state prediction model, thereby obtaining the component health. The decision module 304 is used to generate a set of maintenance schemes that characterize the performance of different maintenance strategies based on component health and a multi-objective maintenance decision optimization method based on stochastic simulation.

[0079] Figure 4 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 4 As shown, the electronic device may include: a processor 401, a communication interface 402, a memory 403, and a communication bus 404. The processor 401, communication interface 402, and memory 403 communicate with each other via the communication bus 404. The processor 401 can call logical instructions from the memory 403 to execute an infrastructure degradation simulation and maintenance optimization decision-making method. This method includes: Acquire multi-source data related to the components of the infrastructure; Spatiotemporal correlation will be established between multi-source data and components in the building information model to construct a component-level data fusion map. Based on the data fusion map, a pre-built physical-data dual-driven degradation state prediction model is used to predict the future degradation state of the component and obtain the component health. Based on component health, a multi-objective maintenance decision optimization method based on stochastic simulation is adopted to generate a set of maintenance schemes that characterize the performance of different maintenance strategies.

[0080] Furthermore, the logical instructions in the aforementioned memory 403 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0081] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is able to execute the infrastructure degradation simulation and maintenance optimization decision-making method provided by the above methods, the method including: Acquire multi-source data related to the components of the infrastructure; Spatiotemporal correlation will be established between multi-source data and components in the building information model to construct a component-level data fusion map. Based on the data fusion map, a pre-built physical-data dual-driven degradation state prediction model is used to predict the future degradation state of the component and obtain the component health. Based on component health, a multi-objective maintenance decision optimization method based on stochastic simulation is adopted to generate a set of maintenance schemes that characterize the performance of different maintenance strategies.

[0082] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the infrastructure degradation simulation and maintenance optimization decision-making methods provided by the above methods, the method comprising: Acquire multi-source data related to the components of the infrastructure; Spatiotemporal correlation will be established between multi-source data and components in the building information model to construct a component-level data fusion map. Based on the data fusion map, a pre-built physical-data dual-driven degradation state prediction model is used to predict the future degradation state of the component and obtain the component health. Based on component health, a multi-objective maintenance decision optimization method based on stochastic simulation is adopted to generate a set of maintenance schemes that characterize the performance of different maintenance strategies.

[0083] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0084] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0085] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for infrastructure deterioration simulation and maintenance optimization decision-making, characterized in that, include: Acquire multi-source data related to the components of the infrastructure; The data will be spatiotemporally correlated with the components in the building information model to construct a component-level data fusion map. Based on the data fusion map, a pre-built physical-data dual-driven degradation state prediction model is used to predict the future degradation state of the component and obtain the component health. Based on the health status of the components, a multi-objective maintenance decision optimization method based on stochastic simulation is used to generate a set of maintenance schemes that characterize the performance of different maintenance strategies.

2. The infrastructure deterioration simulation and maintenance optimization decision-making method according to claim 1, characterized in that, The multi-source data includes real-time monitoring data, historical inspection data, and maintenance history data.

3. The infrastructure deterioration simulation and maintenance optimization decision-making method according to claim 1, characterized in that, The data will be spatiotemporally correlated with the components in the building information model to construct a component-level data fusion map, including: The building information model is then subjected to lightweight processing and semantic enhancement. Assign a unique identifier to the component, expand the semantic attribute slots of the building information model, and explicitly define the data carrying dimensions; The multi-source data is anchored to the component using timestamps and spatial coordinates to generate the data fusion map.

4. The infrastructure deterioration simulation and maintenance optimization decision-making method according to claim 2, characterized in that, Construct a physical-data-driven model for predicting degradation states, including: A physical mechanism model is constructed using damage mechanics as the core framework. The physical mechanism model is subjected to dynamic data calibration and dynamic state calibration to obtain a physical-data dual-driven degradation state prediction model.

5. The infrastructure deterioration simulation and maintenance optimization decision-making method according to claim 4, characterized in that, The dynamic data calibration is achieved by constructing a dynamic parameter correction function; The parameter dynamic correction function uses the real-time monitoring data from the multi-source data to dynamically adjust the model parameters in the physical mechanism model.

6. The infrastructure deterioration simulation and maintenance optimization decision-making method according to claim 4, characterized in that, The dynamic state calibration is achieved by introducing an unscented Kalman filter algorithm; The unscented Kalman filter algorithm includes: Sigma point sampling is used to extract sample points that can accurately represent the probability distribution of the model state of the physical mechanism model; The sample points are weighted and summed to integrate the scattered sample point prediction values ​​into a single state prediction value. The predicted state is corrected by dynamically adjusting the correction weights using Kalman gain.

7. The infrastructure deterioration simulation and maintenance optimization decision-making method according to claim 1, characterized in that, Based on the component health status, a multi-objective maintenance decision optimization method based on stochastic simulation is used to generate a set of maintenance schemes characterizing the performance of different maintenance strategies, including: The Monte Carlo simulation method is used to define the maintenance strategy space and to simulate and deduce multiple preset maintenance strategies. Construct a multi-objective optimization function with both lifecycle cost and safety risk as optimization objectives; The maintenance scheme set is determined by a multi-objective optimization algorithm; wherein the maintenance scheme set is a Pareto optimal scheme set.

8. The infrastructure deterioration simulation and maintenance optimization decision-making method according to claim 1, characterized in that, The method further includes: feeding back the real-time monitoring data after maintenance to the data fusion map and updating the degradation state prediction model.

9. An infrastructure deterioration simulation and maintenance optimization decision-making device, characterized in that, include: The acquisition module is used to acquire multi-source data related to the components of the infrastructure; The association module is used to spatiotemporally associate the multi-source data with the components in the building information model to construct a component-level data fusion map; The prediction module is used to predict the future degradation state of the component based on the data fusion map and using a pre-built physical-data dual-driven degradation state prediction model, so as to obtain the component health. The decision module is used to generate a set of maintenance schemes that characterize the performance of different maintenance strategies based on the health status of the components and a multi-objective maintenance decision optimization method based on stochastic simulation.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the infrastructure degradation simulation and maintenance optimization decision-making method as described in any one of claims 1 to 8.

Images