Method and system for constructing and evolving digital twin of safety state of existing building structure

Abstract

This invention discloses a method and system for constructing and evolving a digital twin of the structural safety status of existing buildings. The method includes encoding key components of the BIM model, compiling monitoring tasks and linking them to the components, and generating an initial digital twin framework with identification and monitoring tasks. Multi-dimensional time-series monitoring data is periodically collected through a sensor network and transmitted to a server in real time. A data assimilation algorithm engine is established, using finite element model parameters as state vectors and monitoring data as observations. The physical state parameters of the components are iteratively calculated and corrected using an unscented Kalman filter algorithm, achieving synchronous evolution between the digital twin and the physical structure. Real-time safety indicators of the components are calculated, compared with multi-level thresholds, and then graded and color-coded on the lightweight BIM model with early warnings issued. Simulated future extreme load conditions are applied to achieve forward-looking risk assessment of the building structure. This provides accurate and efficient quantitative decision support for the intelligent operation and maintenance of the structural safety of existing buildings.

Description

Technical Field

[0001] This invention belongs to the field of information technology in civil engineering and building operation and maintenance management, and more specifically, relates to a method and system for constructing and evolving a digital twin of the safety status of existing building structures. Background Technology

[0002] A large number of existing buildings in my country have entered their middle and old age of use. The safety status of these buildings directly affects the stability of the urban public safety system and is a core research and application concern in the field of building operation and maintenance management. Currently, the safety assessment of existing building structures mainly relies on periodic manual inspections, with a typical inspection cycle of 1-3 years. Furthermore, the assessment conclusions largely depend on the experience and judgment of technical personnel. This traditional assessment model has several insurmountable technical limitations: First, it suffers from a severe lack of timeliness, failing to capture in real time the continuous degradation process of building structure performance under environmental erosion, natural material aging, disturbances from surrounding construction projects, or accidental loads. This easily creates time blind spots in structural safety monitoring, making it difficult to detect potential safety hazards in a timely manner. Second, data management is isolated and fragmented. Assessment reports generated from manual inspections are mostly paper documents or discrete electronic documents, lacking... First, it is difficult to establish an effective link between the method and the building's three-dimensional information model to construct a structural health record for the entire building life cycle, resulting in the inability to achieve collaborative analysis and comprehensive utilization of historical and real-time monitoring data. Second, it lacks proactive early warning and risk prediction capabilities. Traditional methods only focus on the static assessment of the current safety status of the structure and cannot predict the structural mechanical response under future extreme conditions based on real-time monitoring data, making it difficult to achieve early prevention and proactive avoidance of safety accidents. Third, the quantification and intuitiveness of decision support are insufficient. Building operation and maintenance managers find it difficult to obtain a quantitative overall structural safety status, which is not conducive to formulating accurate preventive maintenance, reinforcement, or emergency response plans.

[0003] In recent years, Building Information Modeling (BIM) technology and Internet of Things (IoT) monitoring technology have been increasingly applied in the construction engineering field. However, in actual application scenarios of existing building operation and maintenance, the two technologies have always been used in isolation: BIM models are mostly used as static repositories of geometric and attribute information, reflecting only the design and completion status of the building structure, and unable to reflect the dynamic mechanical state of the structure during use in real time; while IoT monitoring data is displayed independently in the form of lists, charts, etc., lacking a precise and automated association mechanism with specific structural components in the BIM model. Existing technologies have not yet formed effective means to automatically and accurately dynamically bind real-time IoT monitoring data with structural components in the BIM model, nor can they use monitoring data to continuously correct and iteratively update the mechanical parameters embedded in the BIM model. As a result, the BIM model cannot truly reflect the time-varying physical state of the building structure, making it difficult to support model-based real-time structural safety assessment, risk prediction, and proactive early warning.

[0004] In summary, existing technologies for structural safety management of existing buildings have not yet solved the problem of deep integration between static BIM models and dynamic IoT monitoring data. They are unable to construct a data-driven digital twin that can dynamically evolve and accurately reflect the true mechanical state of the physical structure. This has become a key technological bottleneck restricting the improvement of intelligent operation and maintenance levels for existing buildings. How to achieve deep integration of BIM models, IoT monitoring, and mechanical parameter correction to construct a dynamically evolving digital twin of the structural safety status of existing buildings, and realize real-time monitoring, intelligent assessment, and proactive early warning of structural safety, is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] To address the aforementioned deficiencies or improvement needs of existing technologies, this invention provides a method and system for constructing and evolving a digital twin of the safety status of existing building structures. It deeply integrates BIM component uniqueness coding, IoT real-time monitoring, and an unscented Kalman filter data assimilation algorithm. First, component coding enables precise connection between monitoring tasks and key components of the BIM model, generating a finite element reference mechanical model and constructing an initial digital twin framework. Then, multi-dimensional time-series monitoring data is collected and transmitted through a sensor network. Relying on a data assimilation algorithm engine, the physical state parameters of the digital twin are iteratively corrected using the monitoring data as observations, achieving synchronous evolution of the mechanical state of the twin and the solid structure. Finally, based on the evolved twin, the system completes the quantitative calculation of component safety indicators, multi-level threshold comparison, and lightweight BIM model hierarchical visualization annotation and early warning. It can also simulate future extreme load conditions to conduct forward-looking risk assessment of building components. The corresponding system includes model coding and monitoring task connection, monitoring data acquisition and transmission, data assimilation, and dynamic evolution calculation of the digital twin. The invention comprises four modules: calculation, safety assessment, and visual early warning, which work collaboratively through preset interfaces. It endows static BIM models with dynamic evolution capabilities, constructing an intelligent operation and maintenance closed loop of "perception-analysis-decision," achieving a leap from periodic passive assessment to continuous proactive early warning, and from current status judgment to forward-looking risk assessment. Simultaneously, it transforms complex mechanical states into intuitive three-dimensional visual information, significantly reducing the cognitive threshold for operation and maintenance management. This invention enables deep integration of BIM models, IoT monitoring, and mechanical parameter correction, constructing a dynamically evolving digital twin of the existing building structure's structural safety status, achieving real-time monitoring, intelligent assessment, and forward-looking early warning of structural safety. It solves the technical problems of existing technologies, such as the isolation between BIM models and monitoring data, the inability of static models to reflect the true time-varying mechanical state of the structure, lagging safety assessments lacking forward-looking early warning, and a lack of quantitative support for operation and maintenance decisions. It provides accurate and efficient quantitative decision support for the intelligent operation and maintenance of structural safety in existing buildings, especially those in the middle and old age areas.

[0006] To achieve the above objectives, one aspect of the present invention provides a method for constructing and evolving a digital twin of the safety status of an existing building structure, comprising the following steps: S1: Encode each key structural component in the engineering BIM model of the existing building according to the predefined BIM model component unique coding standard; based on the structural safety risk assessment results, compile monitoring task items that include monitored physical quantities, sensor information, and early warning thresholds; associate the monitoring task items with the corresponding components in the BIM model through component coding, and generate a parametric finite element reference mechanical model based on the geometric and material information of the BIM model; combine component coding, monitoring task association relationships, and finite element reference mechanical model to construct an initial digital twin framework with identification and monitoring tasks; S2. Periodically acquire multi-dimensional time-series monitoring data reflecting the structural status by deploying a sensor network at key structural parts of existing buildings and binding it to monitoring tasks, and transmit the monitoring data to the central processing server in real time through a wireless communication protocol. S3. Establish a data assimilation algorithm engine, using the finite element reference mechanical model and its parameters initialized in step S1 as the state vector, and the real-time monitoring data obtained in step S2 as the observation value; perform iterative calculations through the unscented Kalman filter algorithm to obtain the posterior optimal estimate of the state vector, thereby dynamically correcting the physical state parameters of the corresponding components in the digital twin, and driving the finite element reference mechanical model to recalculate, realizing the synchronous evolution of the mechanical state of the digital twin and the solid structure. S4. Calculate the real-time safety index of each component based on the updated digital twin state after step S3; compare the real-time safety index with the preset multi-level safety threshold, and determine the safety level of each component according to the numerical range; perform lightweight processing on the BIM model, and based on the safety level determination results, mark the corresponding components with graded colors on the lightweight 3D BIM model and display the warning information in real time; use the updated digital twin to apply simulated future extreme load conditions, perform structural response prediction, and realize the forward-looking risk assessment of the building structure.

[0007] Furthermore, in step S1, each key structural component in the engineering BIM model of the existing building is coded according to a predefined BIM model component uniqueness coding standard; specifically, this includes the following steps: For existing building engineering BIM models, perform structural component hierarchical processing, with hierarchical dimensions covering at least project, building, floor, and component type; Based on the classification results of structural components, a unique coding rule system for BIM model components is established, which includes classification codes at each level and unique serial numbers of the components. According to the coding rule system, a globally unique identifier is generated for each key structural component in the engineering BIM model, and the coding operation for all key structural components is completed. The initialization digital twin framework The expression is: ; As the mechanical core of the digital twin; its set of parameters to be corrected includes at least the material elastic modulus. That is, the set of parameters to be corrected ; This is a mapping function for the encoding-component-task connection; For the first A key structural component, A collection of components; It is a unique code; The sensor information for the monitoring task item mentioned in step S1 includes at least the sensor type, sensor deployment location, and sensor sampling frequency.

[0008] Furthermore, the monitoring data in step S2 includes at least one or more of strain, displacement, tilt angle, and vibration acceleration; The wireless communication protocol mentioned in step S2 is one or a combination of LoRa, 5G, and WiFi.

[0009] Furthermore, step S3 includes the construction and parameter initialization of the data assimilation algorithm engine, real-time access of model state prediction and monitoring observations, iterative calculation of unscented Kalman filtering, dynamic correction of physical state parameters of the corresponding components of the digital twin, and synchronous evolution of the mechanical state of the digital twin and the solid structure. The data assimilation algorithm engine setup and parameter initialization include: Based on the requirements for nonlinear system parameter estimation, a data assimilation algorithm engine with an embedded unscented Kalman filter algorithm is built, and the core calculation parameters of the algorithm, including process noise covariance, observation noise covariance, and unscented transformation sampling parameters, are configured. The parameterized finite element reference mechanical model generated in step S1 is integrated into the algorithm engine. The physical state parameters to be corrected in the model are integrated to form a model state vector. The dimensions of the state vector, the physical meaning of the parameters corresponding to each dimension, and the initial values ​​are clarified. Configure a data interaction interface for the algorithm engine to enable real-time communication with the central processing server. The engine can automatically and in real-time acquire the monitoring data transmitted in step S2. At the same time, configure a model parameter output interface to support the synchronous push of corrected parameters to the digital twin. Real-time access to model state prediction and monitoring observations includes: The algorithm engine is based on the model state vector at the current moment. By performing structural mechanics simulation calculations using the finite element baseline mechanical model, the theoretical values ​​of the monitored physical quantities of key structural components of existing buildings under current working conditions are predicted, forming a vector of model prediction values. The algorithm engine retrieves the raw monitoring data within the same collection period from the central processing server in real time through the data interaction interface, performs format conversion and normalization on the data, removes invalid interference data, forms an observation value vector that perfectly matches the dimension and physical quantity of the model prediction value vector, and adds a timestamp and corresponding component code to the observation value vector; A "predicted value - observed value" pairing and verification mechanism is established. If there are problems such as mismatch of physical quantities, incorrect component codes, or inconsistent data dimensions, the engine will automatically trigger an alarm and suspend calculation, while simultaneously sending data anomaly information back to the server. The iterative calculation of the unscented Kalman filter specifically includes: Perform an unscented transformation on the optimal model state vector and error covariance at the current moment to generate several sigma sampling points; at the same time, calculate the weight coefficient of each sampling point; Substituting all sigma sampling points into the finite element baseline mechanical model, the predicted sigma sampling points corresponding to each sampling point are obtained through mechanical simulation for state prediction calculation. Then, the predicted sigma sampling points are weighted and summed based on weight coefficients to obtain the prior estimate of the model state vector. and prior error covariance ; For predicted sigma sampling points Perform observation transformation to obtain predicted observations. The prior observation estimates are obtained by weighted summation. Based on the predicted sigma sampling points Predicted observations Prior observation estimates and observation noise covariance matrix Calculate the prior error covariance of the observations and cross covariance Based on the prior error covariance of the observed values and cross covariance Calculate the Kalman gain matrix : ; Combined with real-time monitoring observation vector Prior estimates of the model state vector Make corrections to obtain Posterior optimal state estimation at time 1 ( Simultaneously update the posterior error covariance matrix. ( The posterior optimal estimate is the optimal solution for the model parameters at the current time, and the posterior covariance matrix reflects the accuracy of the parameter estimation. Dynamic correction of the physical state parameters of components corresponding to the digital twin includes: The algorithm engine splits the posterior optimal estimate by component code to obtain the correction value of the physical state parameter to be corrected for each key structural component. The correction value includes the change amount, change trend and current optimal value of the parameter. Following the "one-to-one" component coding matching principle, the split parameter correction values ​​are pushed to the digital twin system to update the physical state parameters of the corresponding components in the twin in real time; the formula for correcting the physical state parameters of the components is: Update ; The synchronous evolution of the mechanical state of the digital twin and the physical structure includes: Substitute the corrected physical state parameters of all components into the finite element reference mechanical model of the digital twin, perform a full model mechanical recalculation, and calculate the real-time mechanical response of each component and part in the digital twin; the updated finite element reference mechanical model for: ;Model As the updated mechanical core of the digital twin, it drives the entire twin framework. Dynamic evolution: ,in, That is A digital twin that evolves in real time, whose mechanical state is synchronously matched with the real state of the physical entity structure; The recalculated mechanical state is compared with that before the recalculation. The impact of parameter correction on the overall mechanical properties of the twin is analyzed to ensure that the changes in the mechanical state are consistent with the actual situation such as the performance degradation of the solid structure and the action of external loads.

[0010] Furthermore, step S3 also includes: after the algorithm engine completes one iteration of calculation and model evolution, it automatically enters the next calculation cycle; repeatedly executes the steps of model state prediction and real-time access of monitoring observations, unscented Kalman filter iterative calculation, dynamic correction of physical state parameters of the corresponding components of the digital twin, and synchronous evolution of the mechanical state of the digital twin and the solid structure, so as to realize the continuous and dynamic evolution of the digital twin with the frequency of monitoring data acquisition cycle; The physical state parameters mentioned in step S3 include at least one of material properties, boundary conditions, and internal force distribution.

[0011] Further, step S4 includes: S41. Extract the full mechanical state data after mechanical recalculation from the digital twin evolved and updated in step S3, and calculate the quantitative real-time safety indicators of each key structural component based on the building structure design code. S42. Compare the calculated real-time safety indicators with the preset multi-level safety thresholds component by component, and determine the safety level of each component according to the value range. S43. Lightweight processing of the original BIM model, and precise color labeling of each component based on the safety level determination results; S44: Automatically generate standardized early warning information for components that are determined to be at the warning level or alarm level, and release it in real time through multiple terminals and in multiple forms; S45: Utilizing an evolved and updated digital twin, simulate the application of future extreme load conditions, calculate changes in structural mechanical response and safety indicators, and achieve a forward-looking assessment of future safety risks of building structures.

[0012] Furthermore, the full mechanical state data mentioned in step S41 includes the stress, displacement, internal force, tilt angle, and vibration characteristics of each key structural component. The data is accompanied by component codes, calculation timestamps, and corresponding load condition information to form a standardized mechanical state dataset. Step S41 also includes binding the calculated real-time security indicators with the corresponding component codes to form a standardized security assessment dataset of "component code - security indicator - calculation time". The quantitative real-time safety indicators of each key structural component in step S41 include one or more of the following: stress ratio, displacement ratio, tilt angle change rate, and vibration amplitude ratio. Step S42 includes: Retrieve the pre-set multi-level security threshold system in the system; An automated comparison method is adopted for each component. The safety index value of each component in the safety assessment dataset is matched with the corresponding threshold intervals. The safety level of the component is determined according to the principle of "value return and level correspondence", and is divided into at least four levels: safety level, attention level, warning level and alarm level. Safety level: index value is below the low threshold; attention level: index value is between the low threshold and the medium threshold; warning level: index value is between the medium threshold and the high threshold; alarm level: index value is above the high threshold. If a component has multiple safety indicators, the final safety level is determined by the principle of comprehensive judgment of multiple indicators, that is, the highest risk level among the corresponding levels of each indicator is taken as the safety level of the component. By associating and integrating component codes, real-time security indicators, threshold comparison results, and final security levels, a "component-indicator-level" judgment dataset is formed.

[0013] Further, step S43 includes: Establish a one-to-one mapping rule between security levels and color codes: green for security level, yellow for alert level, orange for warning level, and red for alarm level. Based on the correspondence between "component code - safety level" in the judgment dataset, precise color annotation at the component level is performed in the lightweight BIM model. The annotation process is automated through code matching. That is, the system locates the corresponding component in the model according to the component code and automatically assigns the color label of the safety level to the component. It supports individual component annotation and batch annotation of the entire model. A visual display interface is built to render and display the lightweight BIM model with graded color annotations in 3D. Picking any component can display its detailed information in real time, achieving a dual display of "visual annotation + refined information". Step S44 includes: Establish rules for automatically generating early warning information. Generate standardized early warning information of corresponding levels based on the safety level of components. The information includes core elements: early warning time, unique component code, spatial location of component, safety index exceeding the standard, safety level, and preliminary handling suggestions. The information content of early warning level and alarm level is differentiated in terms of detail. Alarm level information adds emergency handling prompts. Set the trigger conditions for issuing early warning information: When the safety level of a component is determined to be at the early warning level or alarm level, the system will automatically trigger the issuance process without manual intervention; if the component is upgraded from a low level to a high level, a secondary early warning will also be triggered to remind managers to pay attention to the changes in risk.

[0014] Further, step S45 includes: Establish an extreme load case library containing various extreme load cases that building structures may face, including at least strong winds, earthquakes, nearby construction disturbances, and accidental loads, with standardized load parameters set for each case. Using the current evolved and updated state of the digital twin as the initial condition, single extreme load conditions or combined extreme load conditions from the load case library are applied to the finite element mechanical model of the digital twin according to the building structural mechanics analysis standard. A full-model mechanical response simulation calculation is performed on the digital twin after applying extreme loads to obtain the maximum stress, maximum displacement, and internal force redistribution mechanical parameters of each key structural component under extreme loads. Then, the corresponding safety index is calculated, and the safety level of each component under extreme loads is determined. Based on simulation calculation results, a forward-looking risk assessment is conducted to analyze high-risk components, weak points and possible failure modes of building structures under extreme loads, and a risk assessment report is generated. The report includes working condition parameters, simulation results, a list of high-risk components, risk level assessment and preventive measures recommendations.

[0015] A second aspect of the present invention provides a system for constructing and evolving a digital twin of the safety status of existing building structures, used to implement the method for constructing and evolving a digital twin of the safety status of existing building structures, comprising: The model coding and monitoring task linking module is used to code each key structural component in the engineering BIM model of an existing building according to a predefined BIM model component unique coding standard; based on the structural safety risk assessment results, it compiles monitoring task items that include monitored physical quantities, sensor information, and early warning thresholds; through component coding, it links the monitoring task items with the corresponding components in the BIM model to form an initial digital twin framework with identification and monitoring tasks; at the same time, based on the geometric and material information of the BIM model, it generates a parametric finite element reference mechanical model, which serves as the mechanical core of the digital twin; The monitoring data acquisition and transmission module is used to periodically acquire multi-dimensional time-series monitoring data reflecting the structural status through a sensor network deployed at key structural parts of existing buildings and bound to monitoring tasks, and transmit the monitoring data to the central processing server in real time through a wireless communication protocol. The data assimilation and digital twin dynamic evolution calculation module is used to establish a data assimilation algorithm engine. It uses the finite element reference mechanical model and its parameters initialized in step S1 as the state vector and the real-time monitoring data obtained in step S2 as the observation value. It performs iterative calculations through the unscented Kalman filter algorithm to obtain the posterior optimal estimate of the state vector, thereby dynamically correcting the physical state parameters (material properties, boundary conditions, internal force distribution, etc.) of the corresponding components in the digital twin and driving the finite element reference mechanical model to recalculate, realizing the synchronous evolution of the mechanical state of the digital twin and the solid structure. The safety assessment and visualization early warning module is used to calculate the real-time safety indicators of each component based on the updated digital twin status in step S3; compare the real-time safety indicators with preset multi-level safety thresholds, and determine the safety level of each component according to the numerical range; based on the safety level determination results, classify and color-code the corresponding components on the lightweight 3D BIM model, and display early warning information in real time; and use the updated digital twin to apply simulated future extreme load conditions to predict the structural response, thereby achieving a forward-looking risk assessment of the building structure.

[0016] In summary, compared with the prior art, the above-described technical solutions conceived by this invention can achieve the following beneficial effects: (1) At the level of data fusion and model evolution, this invention achieves precise connection between monitoring tasks and components by constructing a unique coding system for BIM components. Combined with the data assimilation algorithm of unscented Kalman filtering, the static BIM model and the real-time monitoring data of the Internet of Things are deeply integrated, giving the digital twin the ability to dynamically correct physical state parameters. This solves the technical problem in the prior art that the model and monitoring data are isolated and cannot reflect the time-varying mechanical state of the structure, so that the digital twin can achieve synchronous evolution of mechanical state with the physical structure.

[0017] (2) In terms of safety assessment and risk warning, this invention realizes the quantitative calculation of real-time safety indicators of structural components based on dynamic evolution digital twins. Through multi-level threshold comparison and hierarchical visualization labeling of lightweight BIM models, it realizes the intuitive display and real-time warning of structural safety status. At the same time, it can apply simulated future extreme load conditions to carry out structural response prediction, breaking through the limitation of traditional methods that can only conduct current status assessment. It realizes the leap from static assessment to dynamic monitoring, from current status judgment to forward-looking risk assessment, and improves the accuracy and initiative of safety prevention and control of existing building structures.

[0018] At the level of operation and maintenance management and decision support, this invention constructs an intelligent closed-loop operation and maintenance system of "perception-analysis-decision", which transforms the complex structural mechanical state and safety early warning logic into intuitive three-dimensional visualization information, greatly reducing the cognitive threshold for non-professional management personnel. At the same time, it provides quantitative mechanical data and simulation analysis results for the preventive maintenance, reinforcement and renovation and emergency response of existing buildings, solves the problem of insufficient decision support in the traditional management model, and provides efficient and accurate technical support and decision basis for the intelligent operation and maintenance of the structural safety of existing buildings. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating a method for constructing and evolving a digital twin of the safety status of an existing building structure, according to an embodiment of the present invention. Figure 2 This is a schematic diagram illustrating the complete process of constructing and evolving a digital twin of the safety status of an existing building structure according to an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the principle of the dynamic evolution process of the digital twin based on the data assimilation algorithm in a method for constructing and evolving a digital twin of the safety status of an existing building structure according to an embodiment of the present invention. Figure 4 This is a schematic diagram of a system for constructing and evolving a digital twin of the safety status of an existing building structure, according to an embodiment of the present invention. Figure 5 This is a schematic diagram of the four-layer architecture of a digital twin system for constructing and evolving the safety status of existing building structures, according to an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0021] like Figure 1As shown, one aspect of the present invention provides a method for constructing and evolving a digital twin of the safety status of existing building structures. The core objective is to establish a precise coding connection between BIM components and monitoring tasks, and to introduce a data assimilation algorithm to transform the static BIM model and its embedded mechanical model into a digital twin that can be driven by real-time sensing data and dynamically updated to update its physical state parameters. Ultimately, this achieves real-time visual monitoring, intelligent quantitative assessment, and forward-looking risk warning of the safety status of existing building structures, providing precise and efficient technical support for the intelligent operation and maintenance of existing buildings. The method includes the following steps: S1: Model Coding and Monitoring Task Linking: Each key structural component in the engineering BIM model of the existing building is coded according to a predefined BIM model component uniqueness coding standard; based on the structural safety risk assessment results, monitoring task items containing monitoring physical quantities, sensor information, and early warning thresholds are compiled; through component coding, the monitoring task items are associated with the corresponding components in the BIM model, and a parametric finite element reference mechanical model is generated based on the geometric and material information of the BIM model; combining component coding, monitoring task linking relationships, and the finite element reference mechanical model, an initial digital twin framework with identification and monitoring tasks is formed; S2. Monitoring data acquisition and transmission: Multi-dimensional time-series monitoring data reflecting the structural status is periodically acquired by a sensor network deployed in key structural parts of existing buildings and bound to monitoring tasks, and the monitoring data is transmitted to the central processing server in real time through a wireless communication protocol. S3. Dynamic Evolution of Digital Twin Based on Data Assimilation: A data assimilation algorithm engine is established, using the finite element reference mechanical model and its parameters initialized in step S1 as the state vector, and the real-time monitoring data obtained in step S2 as the observation value; the unscented Kalman filter algorithm is used for iterative calculation to obtain the posterior optimal estimate of the state vector, thereby dynamically correcting the physical state parameters of the corresponding components in the digital twin and driving the finite element reference mechanical model to be recalculated, realizing the synchronous evolution of the mechanical state of the digital twin and the solid structure. S4. Safety Assessment and Visual Early Warning: The BIM model is lightweighted, and the real-time safety indicators of each component are calculated based on the updated digital twin state in step S3. The real-time safety indicators are compared with preset multi-level safety thresholds, and the safety level of each component is determined according to the numerical range. Based on the safety level determination results, the corresponding components are graded and color-coded on the lightweighted 3D BIM model, and early warning information is displayed in real time. Using the updated digital twin, simulated future extreme load conditions are applied to predict the structural response and realize the forward-looking risk assessment of the building structure.

[0022] Furthermore, Figure 2The entire process of constructing and evolving a digital twin of the safety status of existing building structures is clearly demonstrated: Based on the BIM model and component coding table, the model coding and monitoring tasks are first linked to complete the initial framework of the digital twin; then, real-time monitoring data collected by sensors is acquired to provide data support for model evolution; next, the digital twin is dynamically updated through a data assimilation algorithm to keep it synchronized with the status of the physical structure; finally, safety assessment and visual early warning are carried out based on the evolved model, and the safety assessment results and early warning model are output, realizing a closed loop from model initialization to operation and maintenance decision support. Furthermore, in step S1, each key structural component in the engineering BIM model of the existing building is coded according to a predefined BIM model component uniqueness coding standard; specifically, this includes the following steps: Key structural components in the engineering BIM model of existing buildings are classified, with the classification dimensions at least covering the project, building, floor, and component type; the classification process meets the following requirements. ,in, A hierarchical set of structural components. Representing the project level, Represents the single-unit level. Represents floor level Represents the component type level, i.e. For the first A key structural component, Each component set corresponds to a unique hierarchical combination. ,in, For the first Each component corresponds to the classification result of level j; the set of classification codes for each level is as follows: ,in ( )correspond Category code; Based on the above classification results of structural components, a unique coding rule system for BIM model components, including classification codes at each level and unique serial numbers for each component, is established; that is... ( ),in A function to generate hierarchical codes; forming a system containing and serial number Encoding rules; According to this coding rule system, a globally unique identifier is generated for each key structural component in the engineering BIM model using a coding function, as shown in the formula: ;in, Each component is uniquely coded to satisfy the uniqueness constraint (i.e., ... ,have ), Complete the coding operation for all key structural components; the coding rule system generation function is: ; Further, in step S1, the monitoring task items are associated with the corresponding components in the BIM model through component coding. Based on the geometric and material information of the BIM model, a parametric finite element reference mechanical model is generated. Combining the component coding, the monitoring task association relationship, and the finite element reference mechanical model, an initialized digital twin framework with identification and monitoring task is formed; including: Organize the key structural component information of the BIM model that has been coded, establish a one-to-one mapping relationship library between component codes and corresponding components in the BIM model, clarify the spatial location, component type and other identity information of the three-dimensional structural component corresponding to each unique component code, and form the basis for accurate matching between component codes and physical components; Organize and compile the information of the completed monitoring task items, and assign a unique component code identifier to each monitoring task item and the target component to form a code-level association; Based on the aforementioned bidirectional association relationship of component codes, an automated binding operation is performed in the digital twin building platform to precisely bind monitoring task items carrying component codes to key structural components with corresponding codes in the BIM model. This ensures that each BIM component is associated with its own dedicated monitoring task item, achieving precise component-level association between monitoring tasks and physical components. The binding relationship satisfies: , This is a mapping function for the encoding-component-task connection; For monitoring tasks; It is a unique code; Geometric information based on BIM model Materials Information Through the generation function Generate parameterized finite element reference mechanical model ; After the BIM model is connected, attribute weights are assigned and a framework is built. The unique code of each component is embedded as the core identity identifier into the attribute information of each component in the BIM model. At the same time, the bound monitoring task items are incorporated into the finite element reference mechanical model system as monitoring attributes of the components, forming an initial digital twin framework with identifiable component identities and traceable monitoring tasks. The initialization expression for the digital twin framework is: ; As the mechanical core of the digital twin; its set of parameters to be corrected includes at least the material elastic modulus. That is, the set of parameters to be corrected ; The sensor information for the monitoring task item mentioned in step S1 includes at least the sensor type, sensor deployment location, and sensor sampling frequency.

[0023] Furthermore, step S2 specifically includes: Based on the association between the monitoring task items that have been attached in step S1 and the key structural components of the BIM model, the physical deployment of the sensor network is completed in the actual key structural parts of the existing building. After the sensor network is deployed and debugged, the status data of the key structures of the existing building are automatically and periodically collected according to the sampling frequency preset in the monitoring task items. The multi-dimensional time-series monitoring data collected by the sensor network is packaged and transmitted in real time through wireless communication protocols, and finally sent to the central processing server. The monitoring data mentioned in step S2 includes at least one or more of strain, displacement, tilt angle, and vibration acceleration. The data must be accompanied by a timestamp of acquisition, a unique sensor identifier, and the corresponding component code, forming a three-dimensional data label of "time-data-component". The acquisition cycle strictly follows the preset standards of the monitoring task (e.g., 1 time / minute, 5 times / minute, etc.) to ensure the continuity and regularity of data acquisition, forming a complete time-series data sequence that reflects the dynamic changes in the structural state. The acquisition process is automatically executed by the sensor network without manual intervention and also has a self-check function for data acquisition anomalies. If acquisition interruption or data loss occurs, a self-check can be automatically triggered and abnormal information recorded. The wireless communication protocol mentioned in step S2 is one or a combination of LoRa, 5G, and WiFi.

[0024] Furthermore, step S3 is the core step in realizing the digital twin's transformation from "static initialization" to "dynamic evolution." It involves deeply integrating real-time monitoring data with the mechanical model through a data assimilation algorithm, achieving synchronous matching between the digital twin's mechanical state and the physical structure. Specifically, this includes five sub-steps: building and initializing the data assimilation algorithm engine, real-time access to model state prediction and monitoring observations, iterative calculation using unscented Kalman filtering, dynamic correction of the physical state parameters of the corresponding components in the digital twin, and synchronous evolution of the mechanical states of the digital twin and the physical structure. First, a data assimilation algorithm engine adapted to the mechanical characteristics of the building structure is built, and the core parameters and basic model of the engine are initialized, providing a foundation for subsequent iterative calculations. The algorithm engine executes model state prediction at a preset period (consistent with the monitoring data acquisition period in step S2) and synchronously accesses real-time monitoring data. Using the observed values ​​as the basis, the pairing of "model prediction value - actual observation value" is achieved. Then, based on the model state vector, prediction value vector, and observation value vector, the unscented Kalman filter algorithm is executed for full-process iterative calculation to obtain the posterior optimal estimate of the state vector, thus achieving the optimal solution of the model parameters. Based on the posterior optimal estimate obtained by the algorithm iterative calculation, the physical state parameters of each key structural component in the digital twin are accurately and automatically dynamically corrected to ensure that the parameters are consistent with the real state of the physical structure. Finally, based on the corrected physical state parameters, the overall mechanical model of the digital twin is recalculated to achieve synchronous matching between the mechanical state of the twin and the real mechanical state of the existing building structure, completing one dynamic evolution of the digital twin. The physical state parameters mentioned in step S3 include at least one of material properties, boundary conditions, and internal force distribution.

[0025] Furthermore, the data assimilation algorithm engine setup and parameter initialization include: Based on the requirements for nonlinear system parameter estimation, a data assimilation algorithm engine with an embedded unscented Kalman filter (UKF) algorithm was built, and the core calculation parameters of the algorithm (including process noise covariance, observation noise covariance, unscented transformation sampling parameters, etc.) were configured. The initial values ​​of the parameters were set based on building structure design specifications and engineering measurement experience. The parameterized finite element reference mechanical model generated in step S1 is integrated into the algorithm engine. The physical state parameters to be corrected in the model (including at least the material elastic modulus, and may also include material properties, boundary conditions, internal force distribution, etc.) are integrated to construct the model state vector. Clarify the dimensions of the state vector, the physical meaning of the parameters corresponding to each dimension, and the initial values ​​(the initial values ​​are set based on the geometry and material information of the BIM model). Configure a data interaction interface for the algorithm engine to enable real-time communication with the central processing server, ensuring that the engine can automatically and in real time acquire the monitoring data transmitted in step S2. At the same time, configure a model parameter output interface to support the synchronous push of corrected parameters to the digital twin. Real-time access to model state predictions and monitoring observations, including: The algorithm engine uses the model state vector at the current moment (the optimal parameter set after correction in the previous calculation cycle) to perform structural mechanics simulation calculations through the finite element reference mechanical model. It predicts the theoretical values ​​of the monitored physical quantities of each key structural component of the existing building under the current working conditions, forming a model prediction value vector. The predicted values ​​correspond one-to-one with the monitored physical quantities (strain, displacement, tilt angle, vibration acceleration, etc.) in step S2. The algorithm engine retrieves the raw monitoring data within the same collection period from the central processing server in real time through the data interaction interface, performs format conversion and normalization on the data, removes invalid interference data, and forms a monitoring observation vector that perfectly matches the dimension and physical quantity of the model prediction value vector. It also adds timestamps and corresponding component codes to the monitoring observation vector to ensure accurate matching with the model prediction value vector. A "predicted value - observed value" pairing and verification mechanism is established. If problems such as mismatch of physical quantities, incorrect component coding, or inconsistent data dimensions occur, the engine will automatically trigger an alarm and suspend calculation, while simultaneously sending abnormal data information back to the server. Establishing a "predicted value-observed value" pairing verification mechanism refers to constructing an automated verification and anomaly detection process for the model's predicted value vector and the monitoring observed value vector before the data assimilation algorithm engine performs iterative calculations. This mechanism uses component coding as the unique link to perform bidirectional verification of two types of data within the same time period and under the same monitored physical quantity dimension, including: Dimension and physical quantity matching verification: Verify whether the dimensions and corresponding monitored physical quantities (strain, displacement, tilt angle, vibration acceleration, etc.) of the model predicted value vector and the monitoring observed value vector are completely consistent, ensuring that each predicted value has a unique corresponding observed value, and there are no problems of dimension mismatch or physical quantity mismatch. Component code correspondence verification: Verify whether the unique codes of BIM components bound to the two types of data correspond one-to-one, ensuring that the predicted value and the observed value belong to the same key structural component, and avoid parameter correction errors caused by cross-component data pairing; Data validity verification: Verify whether the observed values ​​are missing, abnormal (such as exceeding the sensor range, excessive abrupt change), or have format errors, and at the same time verify whether the predicted values ​​are within the reasonable range of structural mechanics, and eliminate invalid interference data; Anomaly Handling Logic: If the verification finds any of the above mismatches or anomalies, the mechanism will automatically trigger an alarm and suspend the current cycle of unscented Kalman filter iterative calculation. At the same time, it will send the anomaly information (such as the abnormal component code, abnormal data type, and anomaly cause) back to the central processing server. The calculation process will resume after the data is reviewed and corrected.

[0026] The iterative calculation of the unscented Kalman filter specifically includes the following steps: The iterative calculation of the unscented Kalman filter specifically includes the following steps: Unscented transform sampling: based on Optimal model state vector at time 1 and error covariance Perform an unscented transformation to generate sigma sampling point set , State vector The dimension; the number of sampling points is set based on the dimension of the state vector to ensure coverage of the main space of parameter changes, while calculating the weight coefficient of each sampling point; Sampling point state prediction: Predict all sigma sampling points Substitute the finite element baseline mechanical model F, and perform mechanical simulation. The state prediction calculation at each time point yields the predicted sigma sampling points corresponding to each sampling point. The model state vector is then obtained by weighting and summing the predicted sigma sampling points based on the weighting coefficients. and prior error covariance : in, and The weighting coefficients for the state estimate and the error covariance are respectively, satisfying... and ; The process noise covariance matrix; Kalman gain calculation: for predicted sigma sampling points Perform observation transformation to obtain predicted observations. The prior observation estimates are obtained by weighted summation. : in, For the observation model function, the mechanical state is mapped to a measurable physical quantity; Based on the predicted sigma sampling points Predicted observations Prior observation estimates and observation noise covariance matrix Calculate the prior error covariance of the observations and cross covariance : Based on the prior error covariance of the observed values and cross covariance Calculate the Kalman gain matrix : The value of the Kalman gain determines the weight of the monitoring observations on the correction of the model state vector, realizing the dynamic weight allocation between "model prediction" and "actual monitoring". Posterior optimal estimation solution: combining real-time monitoring observation vectors Prior estimates of the model state vector Make corrections to obtain Posterior optimal state estimation at time 1 ( Simultaneously update the error covariance matrix. ( ); The posterior optimal estimate is the optimal solution of the model parameters at the current time, and the posterior covariance matrix reflects the accuracy of the parameter estimation. Dynamic correction of the physical state parameters of components corresponding to the digital twin includes: The algorithm engine estimates the posterior optimal state. The components are split according to their codes to obtain the corrected physical state parameters for each key structural component. The correction value includes the amount of parameter change, the trend of change, and the current optimal value; Following the "one-to-one" component coding matching principle, the split parameter correction values ​​are pushed to the digital twin system to update the physical state parameters of the corresponding components in the twin in real time. The updated parameters include at least the material elastic modulus, and can also be updated according to engineering needs, such as material strength, boundary constraints, and component internal force distribution. Establish a parameter correction threshold constraint mechanism. If the correction amount of a certain parameter exceeds the preset reasonable range (such as the sudden change of the elastic modulus of the material is too large), the engine will automatically determine it as an abnormal correction, suspend parameter updates and trigger data verification. After eliminating interference factors such as sensor failure and data transmission error, the correction operation will be performed again. The entire process of parameter correction for all components is recorded, including correction time, component code, original parameter value, correction value, and correction basis (corresponding observation data), forming a parameter evolution archive to support subsequent traceability and analysis; The formula for correcting the physical state parameters of a component is: Update ; The synchronous evolution of the mechanical state of the digital twin and the physical structure includes: Substitute the corrected physical state parameters of all components into the finite element reference mechanical model of the digital twin, perform a full model mechanical recalculation, and calculate the real-time mechanical response (including stress, strain, displacement, internal force, vibration characteristics, etc.) of each component and part in the digital twin; the updated finite element reference mechanical model for: This model serves as the updated mechanical core of the digital twin, driving the entire twin framework. The dynamic evolution is expressed by the following formula: ,in, That is A digital twin that evolves in real time, whose mechanical state is synchronously matched with the real state of the physical entity structure; The mechanical state after recalculation is compared with that before the recalculation. The influence of parameter correction on the overall mechanical properties of the twin is analyzed to ensure that the change in mechanical state is consistent with the actual situation such as the performance degradation of the solid structure and the action of external loads. The system completes the synchronous update and storage of the mechanical state of the digital twin, saving the current mechanical state as the latest state of the twin, forming a three-dimensional evolution archive of "time-parameter-mechanical state", and realizing the full tracking of the mechanical state change process of the physical structure. After the algorithm engine completes one iteration of calculation and model evolution, it automatically enters the next calculation cycle. It repeatedly executes the steps of model state prediction and real-time access of monitoring observations, unscented Kalman filter iterative calculation, dynamic correction of physical state parameters of the corresponding components of the digital twin, and synchronous evolution of the mechanical state of the digital twin and the physical structure. This enables the digital twin to continuously and dynamically evolve with the frequency of monitoring data acquisition cycle, ensuring that the twin always remains synchronized with the real state of the physical structure.

[0027] Furthermore, Figure 3 The diagram visually illustrates the dynamic evolutionary closed-loop process of a digital twin based on a data assimilation algorithm: Digital twin state vector at time step Based on, combined Real-time sensor observation data collected at all times The data assimilation algorithm engine performs iterative calculations and parameter optimizations in the intermediate stage, ultimately outputting the updated data. Time-Temporal Digital Twin State Vector This enables the synchronous matching and continuous evolution of the digital twin with the real state of the physical entity structure, forming a complete closed loop of "historical state - real-time observation - algorithm assimilation - state update".

[0028] Furthermore, step S4, based on the dynamically evolved twin state, achieves quantitative assessment, visualization, and forward-looking early warning of structural safety, forming a closed loop from mechanical state analysis to operation and maintenance decision support. Specifically, it is refined into five sub-steps: twin state extraction and safety index calculation, safety index threshold comparison and level determination, lightweight BIM model hierarchical visualization annotation, early warning information generation and multi-terminal release, extreme load condition simulation and forward-looking risk assessment. The hierarchical color annotation levels in step S4 include at least safe, attention, early warning, and alarm, corresponding to green, yellow, orange, and red, respectively. The future extreme load conditions mentioned in step S4 include at least one of strong wind, earthquake, nearby construction disturbance, and accidental load.

[0029] Step S4 specifically includes: S41: Extract full mechanical state data from the digital twin evolved and updated in step S3, and calculate the quantitative real-time safety indicators of each key structural component based on the building structure design code. S42: Compare the calculated real-time safety indicators with the preset multi-level safety thresholds component by component, and determine the safety level of each component according to the numerical range. S43: Lightweight processing of the original BIM model, precise classification and color labeling of each component based on the safety level judgment results, to realize the three-dimensional spatial visualization of the building structure safety status and reduce the cognitive threshold for non-professionals. S44: Automatically generate standardized early warning information for components that are determined to be at the warning or alarm level, and release it in real time through multiple terminals and in multiple forms to ensure that operation and maintenance management personnel can obtain early warning information as soon as possible; S45: Utilizing the evolved and updated digital twin, simulate the application of future extreme load conditions, calculate the changes in structural mechanical response and safety indicators, realize the forward-looking assessment of the future safety risks of building structures, and provide quantitative basis for the formulation of preventive maintenance and reinforcement schemes; Step S41 includes: The full mechanical state data after mechanical recalculation is automatically extracted from the digital twin system, including mechanical state parameters such as stress, displacement, internal force, tilt angle, and vibration characteristics of each key structural component. The data is accompanied by component codes, calculation timestamps, and corresponding load condition information to form a standardized mechanical state dataset. Based on the current building structure design codes such as the "Code for Design of Concrete Structures" and the "Standard for Design of Steel Structures" and the calculation rules set by actual engineering, the mechanical state dataset is analyzed and calculated to obtain the quantitative real-time safety indicators of each key structural component. These indicators include at least the stress ratio (actual stress of the component / allowable design stress) and the displacement ratio (actual displacement of the component / allowable design displacement). Derivative safety indicators such as the tilt angle change rate and vibration amplitude ratio can also be extended according to engineering needs. Establish a safety indicator calculation verification mechanism to automatically detect situations such as missing parameters, incorrect formula application, and excessive values ​​during the calculation process. If a calculation anomaly occurs, immediately backtrack the twin mechanical state data and recalculate, while recording the anomaly information to ensure the accuracy of the safety indicator values. The calculated real-time safety indicators are bound to the corresponding component codes to form a standardized safety assessment dataset of "component code - safety indicator - calculation time", which is then stored in the system database. Step S42 includes: The system retrieves a multi-level safety threshold system pre-set in the system. This threshold system is formulated based on building structure design specifications, engineering expert experience, and structural safety risk assessment results. It corresponds one-to-one with safety indicators (such as stress ratio threshold and displacement ratio threshold), and the thresholds can be flexibly adjusted according to the building's service life and changes in surrounding working conditions. The system employs an automated component-by-component comparison method to match the safety index values ​​of each component in the safety assessment dataset with the corresponding threshold intervals. The safety level of the component is determined according to the principle of "value return and level correspondence," and is divided into at least four levels: safety level (index value is below the low threshold), attention level (index value is between the low threshold and the medium threshold), warning level (index value is between the medium threshold and the high threshold), and alarm level (index value is above the high threshold). Each level has clearly defined numerical interval boundaries, with no ambiguity in the determination. If a component has multiple safety indicators (such as simultaneously calculating stress ratio and displacement ratio), the final safety level is determined by the principle of comprehensive judgment of multiple indicators. That is, the highest risk level among the corresponding levels of each indicator is taken as the safety level of the component, thus avoiding the limitations of judgment by a single indicator. By associating and integrating component codes, real-time security indicators, threshold comparison results, and final security levels, a "component-indicator-level" judgment dataset is formed.

[0030] Step S43 includes: The original 3D BIM model of the digital twin is lightweighted by simplifying the number of model faces, compressing textures, and optimizing component levels. This reduces the amount of model data while ensuring the accuracy of component geometric features and spatial positions, and supports fast loading and rendering on multiple terminals such as Web and mobile devices. Establish a one-to-one mapping rule between security levels and color codes. The rule is an industry-standard and easily recognizable one: green corresponds to security level, yellow corresponds to attention level, orange corresponds to warning level, and red corresponds to alarm level. Each color is assigned a uniform RGB value to ensure the consistency of the labeled colors. Based on the correspondence between "component code - safety level" in the judgment dataset, precise color annotation at the component level is performed in the lightweight BIM model. The annotation process is automated through code matching. That is, the system locates the corresponding component in the model according to the component code and automatically assigns the color label of the safety level to the component. It supports individual component annotation and batch annotation of the entire model. A visual display interface is built to render and display the lightweight BIM model with graded color annotations in 3D. It supports interactive operations such as model rotation, scaling, sectioning, and component picking. Picking any component can display its detailed information in real time, including component code, component type, real-time safety index value, safety level, and the time of the last parameter correction, achieving a dual display of "visual annotation + refined information". Step S44 includes: Establish rules for automatically generating early warning information. Generate standardized early warning information of corresponding levels based on the safety level of components. The information includes core elements: early warning time, unique component code, spatial location of component, safety indicators exceeding standards (value, threshold, over-amplitude ratio), safety level, and preliminary handling suggestions. The information content of early warning level and alarm level is differentiated in terms of detail. Alarm level information adds emergency handling prompts. Set the trigger conditions for issuing early warning information: When the safety level of a component is determined to be at the early warning level or alarm level, the system will automatically trigger the issuance process without manual intervention; if the component is upgraded from a low level to a high level (such as from the attention level to the early warning level), a secondary early warning will also be triggered to remind managers to pay attention to the changes in risk. A multi-terminal collaborative publishing approach is adopted to cover the main information receiving channels for operation and maintenance management, including: pop-up display of the digital twin platform early warning center, message push of the mobile APP of operation and maintenance management personnel, SMS / email reminders, and sound and light alarms in the on-site monitoring room (only for alarm level), to ensure that early warning information is delivered without omission; All early warning information is recorded and archived throughout the entire process, and an early warning information ledger is established. The recorded content includes early warning number, trigger time, component information, early warning level, release channel, receiving personnel, handling status, and closed-loop time. It supports the query, traceability, and statistical analysis of early warning information, and provides data support for subsequent operation and maintenance summaries.

[0031] Step S45 includes: Build an extreme load case library containing various extreme load cases that building structures may face, including at least strong winds, earthquakes, disturbances from nearby construction (such as foundation pit excavation and subway construction), and accidental loads (such as local surcharges and impacts). Standardized load parameters (such as wind speed, seismic intensity, and earth pressure values) are set for each case, and the library supports adding and modifying case parameters according to the actual project. Using the current evolved and updated state of the digital twin as the initial condition, single extreme load conditions or combined extreme load conditions (such as strong wind + earthquake) from the load case library are applied to the finite element mechanical model of the digital twin according to the building structural mechanics analysis standard. A full-model mechanical response simulation calculation is performed on the digital twin after applying extreme loads to obtain the mechanical parameters such as maximum stress, maximum displacement, and internal force redistribution of each key structural component under extreme loads. Then, the corresponding safety index is calculated to determine the safety level of each component under extreme loads. Based on simulation calculation results, a forward-looking risk assessment is conducted to analyze high-risk components, weak points and possible failure modes of building structures under extreme loads, and generate a risk assessment report. The report includes core contents such as working condition parameters, simulation results, a list of high-risk components, risk level assessment, and preventive treatment recommendations (such as reinforcement points and maintenance measures). It supports operation and maintenance personnel to conduct multi-condition comparative simulations based on actual needs, analyze the differences in the safety status of structures under different extreme loads, and provide quantitative simulation data support for formulating targeted risk prevention and control plans and optimizing reinforcement strategies, thereby realizing the upgrade of safety management from "post-event response" to "pre-event prevention".

[0032] like Figure 4 As shown, a second aspect of the present invention provides a system for constructing and evolving a digital twin of the safety status of existing building structures, used to implement the above-mentioned design method, comprising: The model coding and monitoring task linking module is used to code each key structural component in the engineering BIM model of an existing building according to a predefined BIM model component unique coding standard; based on the structural safety risk assessment results, it compiles monitoring task items that include monitored physical quantities, sensor information, and early warning thresholds; through component coding, it links the monitoring task items with the corresponding components in the BIM model to form an initial digital twin framework with identification and monitoring tasks; at the same time, based on the geometric and material information of the BIM model, it generates a parametric finite element reference mechanical model, which serves as the mechanical core of the digital twin; The monitoring data acquisition and transmission module is used to periodically acquire multi-dimensional time-series monitoring data reflecting the structural status through a sensor network deployed at key structural parts of existing buildings and bound to monitoring tasks, and transmit the monitoring data to the central processing server in real time through a wireless communication protocol. The data assimilation and digital twin dynamic evolution calculation module is used to establish a data assimilation algorithm engine. It uses the finite element reference mechanical model and its parameters initialized in step S1 as the state vector and the real-time monitoring data obtained in step S2 as the observation value. It performs iterative calculations through the unscented Kalman filter algorithm to obtain the posterior optimal estimate of the state vector, thereby dynamically correcting the physical state parameters (material properties, boundary conditions, internal force distribution, etc.) of the corresponding components in the digital twin and driving the finite element reference mechanical model to recalculate, realizing the synchronous evolution of the mechanical state of the digital twin and the solid structure. The safety assessment and visualization early warning module is used to perform lightweight processing on the BIM model, calculate the real-time safety indicators of each component based on the updated digital twin state in step S3, compare the real-time safety indicators with preset multi-level safety thresholds, and determine the safety level of each component according to the numerical range; based on the safety level determination results, the corresponding components are graded and color-coded on the lightweight 3D BIM model, and early warning information is displayed in real time; using the updated digital twin, simulated future extreme load conditions are applied to predict the structural response and realize the forward-looking risk assessment of the building structure.

[0033] The four modules mentioned above form a collaborative relationship with one-way data output and two-way information feedback according to the execution logic of method steps S1→S2→S3→S4. The output results of the previous module serve as the execution input of the next module, and each module can feed back abnormal information and data verification results during the execution process to the upstream module. Through the preset data interface, real-time interaction and accurate transmission of data throughout the process are realized, ensuring the orderly execution of method steps and the stable implementation of system functions.

[0034] Furthermore, Figure 5 The system clearly presents the four-layer architecture logic of the digital twin construction and evolution system for the structural safety status of existing buildings: the bottom perception layer realizes bidirectional perception docking between physical entities and digital models through sensor networks and BIM models; the middle data layer relies on model / task databases and real-time monitoring databases to complete data interaction and drive evolution; the upper computing and service layer realizes dynamic parameter correction and status update of the digital twin through data assimilation and evolution computing modules; the top application layer uses a visualization early warning platform as a carrier to output structural safety assessment results and forward-looking early warning information, forming a complete technical closed loop of "perception-data-computation-application", providing full-process support for intelligent operation and maintenance of existing buildings.

[0035] It should be noted that the existing building structure safety status digital twin construction and evolution system provided in this embodiment can be a computer program (including program code) running on a computer device. For example, the existing building structure safety status digital twin construction and evolution system is an application software; the existing building structure safety status digital twin construction and evolution system can be used to execute the corresponding steps in the above-mentioned methods provided in the embodiments of this application.

[0036] This application also provides a computer-readable storage medium storing a computer program that is executed by a processor to implement... Figure 1 The methods provided in each step are detailed in the implementation methods provided in the above steps, and will not be repeated here.

[0037] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for constructing and evolving a digital twin of the safety state of an existing building structure, characterized in that, Includes the following steps: S1: Code each key structural component in the engineering BIM model of the existing building according to the predefined BIM model component unique coding standard; Based on the structural safety risk assessment results, monitoring task items are compiled, including monitored physical quantities, sensor information, and early warning thresholds; By using component coding, monitoring tasks are associated with corresponding components in the BIM model. Based on the geometric and material information of the BIM model, a parametric finite element reference mechanical model is generated. Combining component coding, monitoring task association, and finite element reference mechanical model, an initial digital twin framework with identification and monitoring tasks is constructed. S2. Periodically acquire multi-dimensional time-series monitoring data reflecting the structural status by deploying a sensor network at key structural parts of existing buildings and binding it to monitoring tasks, and transmit the monitoring data to the central processing server in real time through a wireless communication protocol. S3. Establish a data assimilation algorithm engine, using the finite element reference mechanical model and its parameters initialized in step S1 as the state vector, and the real-time monitoring data obtained in step S2 as the observation value. Iterative calculations are performed using the unscented Kalman filter algorithm to obtain the posterior optimal estimate of the state vector, thereby dynamically correcting the physical state parameters of the corresponding components in the digital twin and driving the finite element reference mechanical model to be recalculated, thus realizing the synchronous evolution of the mechanical state of the digital twin and the solid structure. S4. Calculate the real-time security index of each component based on the updated state of the digital twin in step S3; compare the real-time security index with the preset multi-level security threshold, and determine the security level of each component according to the value range. The BIM model is lightweighted. Based on the safety level assessment results, the corresponding components are graded and color-coded on the lightweighted 3D BIM model, and early warning information is displayed in real time. By using the updated digital twin, simulated future extreme load conditions are applied to predict the structural response and achieve forward-looking risk assessment of building structures.

2. The method of claim 1, wherein: In step S1, each key structural component in the engineering BIM model of the existing building is coded according to the predefined BIM model component uniqueness coding standard. Specifically, the following steps are included: For existing building engineering BIM models, perform structural component hierarchical processing, with hierarchical dimensions covering at least project, building, floor, and component type; Based on the classification results of structural components, a unique coding rule system for BIM model components is established, which includes classification codes at each level and unique serial numbers of the components. According to the aforementioned coding rule system, a globally unique identifier is generated for each key structural component in the engineering BIM model; The initialized digital twin framework Expression is: ; The mechanical kernel as a digital twin; its set of parameters to be corrected contains at least the material elastic modulus , i.e. the set of parameters to be corrected ; Is the encoding-component-task hooking mapping function; Is the Key structural component, Is the set of components; Is the unique encoding; The sensor information for the monitoring task item mentioned in step S1 includes at least the sensor type, sensor deployment location, and sensor sampling frequency.

3. The method of claim 2, wherein: the method further comprises: determining a plurality of digital twins of the existing building structure; and determining a plurality of digital twins of the existing building structure based on the plurality of digital twins of the existing building structure. The monitoring data mentioned in step S2 includes at least one or more of strain, displacement, tilt angle, and vibration acceleration; The wireless communication protocol mentioned in step S2 is one or a combination of LoRa, 5G, and WiFi.

4. The method of claim 1-3, wherein: Step S3 includes unscented Kalman filter iterative calculation, dynamic correction of the physical state parameters of the corresponding components of the digital twin, and synchronous evolution of the mechanical state of the digital twin and the solid structure. The iterative calculation of the unscented Kalman filter specifically includes: Perform an unscented transformation on the optimal model state vector and error covariance at the current moment to generate several sigma sampling points; at the same time, calculate the weight coefficient of each sampling point; Substituting all sigma sampling points into the finite element baseline mechanical model, the predicted sigma sampling points corresponding to each sampling point are obtained through mechanical simulation for state prediction calculation. Then, the predicted sigma sampling points are weighted and summed based on weight coefficients to obtain the prior estimate of the model state vector. and prior error covariance ; For predicted sigma sampling points Perform observation transformation to obtain predicted observations. The prior observation estimates are obtained by weighted summation. Based on the predicted sigma sampling points Predicted observations Prior observation estimates and observation noise covariance matrix Calculate the prior error covariance of the observations and cross covariance Based on the prior error covariance of the observed values and cross covariance Calculate the Kalman gain matrix : ; Combined with real-time monitoring observation vector Prior estimates of the model state vector Make corrections to obtain Posterior optimal state estimation at time 1 ( Simultaneously update the posterior error covariance matrix. ( The posterior optimal estimate is the optimal solution for the model parameters at the current time, and the posterior covariance matrix reflects the accuracy of the parameter estimation. Dynamic correction of the physical state parameters of components corresponding to the digital twin includes: The algorithm engine splits the posterior optimal estimate by component code to obtain the correction value of the physical state parameter to be corrected for each key structural component. The correction value includes the change amount, change trend and current optimal value of the parameter. Following the "one-to-one" component coding matching principle, the split parameter correction values ​​are pushed to the digital twin system to update the physical state parameters of the corresponding components in the twin in real time; the formula for correcting the physical state parameters of the components is: Update ; The synchronous evolution of the mechanical state of the digital twin and the physical structure includes: Substitute the corrected physical state parameters of all components into the finite element reference mechanical model of the digital twin, perform a full model mechanical recalculation, and calculate the real-time mechanical response of each component and part in the digital twin; the updated finite element reference mechanical model for: ;Model As the updated mechanical core of the digital twin, it drives the entire twin framework. Dynamic evolution: ,in, That is A digital twin that evolves in real time, whose mechanical state is synchronously matched with the real state of the physical entity structure; The recalculated mechanical state is compared with that before the recalculation. The impact of parameter correction on the overall mechanical properties of the twin is analyzed to ensure that the changes in the mechanical state are consistent with the actual situation such as the performance degradation of the solid structure and the action of external loads.

5. The method for constructing and evolving a digital twin of the safety status of an existing building structure according to claim 4, characterized in that: Step S3 also includes: after the algorithm engine completes one iteration of calculation and model evolution, it automatically enters the next calculation cycle; repeatedly executes the steps of model state prediction and real-time access of monitoring observations, unscented Kalman filter iterative calculation, dynamic correction of physical state parameters of the corresponding components of the digital twin, and synchronous evolution of the mechanical state of the digital twin and the solid structure, so as to realize the continuous and dynamic evolution of the digital twin with the frequency of monitoring data acquisition cycle; The physical state parameters mentioned in step S3 include at least one of material properties, boundary conditions, and internal force distribution.

6. A method for constructing and evolving a digital twin of the safety status of an existing building structure according to any one of claims 1-3, characterized in that: Step S4 includes: S41. Extract the full mechanical state data after mechanical recalculation from the digital twin evolved and updated in step S3, and calculate the quantitative real-time safety indicators of each key structural component based on the building structure design code. S42. Compare the calculated real-time safety indicators with the preset multi-level safety thresholds component by component, and determine the safety level of each component according to the value range. S43. Lightweight processing of the original BIM model, and precise color labeling of each component based on the safety level determination results; S44: Automatically generate standardized early warning information for components that are determined to be at the warning level or alarm level, and release it in real time through multiple terminals and in multiple forms; S45: Utilizing an evolved and updated digital twin, simulate the application of future extreme load conditions, calculate changes in structural mechanical response and safety indicators, and achieve a forward-looking assessment of future safety risks of building structures.

7. The method for constructing and evolving a digital twin of the safety status of an existing building structure according to claim 6, characterized in that: The full mechanical state data mentioned in step S41 includes the stress, displacement, internal force, tilt angle, and vibration characteristics of each key structural component. The data is accompanied by component codes, calculation timestamps, and corresponding load condition information to form a standardized mechanical state dataset. Step S41 also includes binding the calculated real-time security indicators with the corresponding component codes to form a standardized security assessment dataset of "component code - security indicator - calculation time". The quantitative real-time safety indicators of each key structural component in step S41 include one or more of the following: stress ratio, displacement ratio, tilt angle change rate, and vibration amplitude ratio. Step S42 includes: Retrieve the pre-set multi-level security threshold system in the system; An automated comparison method is adopted for each component. The safety index value of each component in the safety assessment dataset is matched with the corresponding threshold intervals. The safety level of the component is determined according to the principle of "value return and level correspondence", and is divided into at least four levels: safety level, attention level, warning level and alarm level. Safety level: index value is below the low threshold; attention level: index value is between the low threshold and the medium threshold; warning level: index value is between the medium threshold and the high threshold; alarm level: index value is above the high threshold. If a component has multiple safety indicators, the final safety level is determined by the principle of comprehensive judgment of multiple indicators, that is, the highest risk level among the corresponding levels of each indicator is taken as the safety level of the component. By associating and integrating component codes, real-time security indicators, threshold comparison results, and final security levels, a "component-indicator-level" judgment dataset is formed.

8. The method for constructing and evolving a digital twin of the safety status of an existing building structure according to claim 7, characterized in that: Step S43 includes: Establish a one-to-one mapping rule between security levels and color codes: green for security level, yellow for alert level, orange for warning level, and red for alarm level. Based on the "component code - safety level" correspondence of the judgment dataset, precise color annotation at the component level is performed in the lightweight BIM model. The annotation process is automated through code matching. That is, the system locates the corresponding component in the model according to the component code and automatically assigns the color label of the safety level to the component. It supports individual component annotation and batch annotation of the entire model. A visual display interface is built to render and display the lightweight BIM model with graded color annotations in 3D. Picking any component can display its detailed information in real time, realizing a dual display of "visual annotation + refined information". Step S44 includes: Establish rules for automatically generating early warning information. Generate standardized early warning information of corresponding levels based on the safety level of components. The information includes core elements: early warning time, unique component code, spatial location of component, safety index exceeding the standard, safety level, and preliminary handling suggestions. The information content of early warning level and alarm level is differentiated in terms of detail. Alarm level information adds emergency handling prompts. Set the trigger conditions for issuing early warning information: When the safety level of a component is determined to be at the early warning level or alarm level, the system will automatically trigger the issuance process without manual intervention; if the component is upgraded from a low level to a high level, a secondary early warning will also be triggered to remind managers to pay attention to the changes in risk.

9. The method for constructing and evolving a digital twin of the structural safety status of an existing building according to claim 8, characterized in that: Step S45 includes: Establish an extreme load case library containing various extreme load cases that building structures may face, including at least strong winds, earthquakes, nearby construction disturbances, and accidental loads, with standardized load parameters set for each case. Using the current evolved and updated state of the digital twin as the initial condition, single extreme load conditions or combined extreme load conditions from the load case library are applied to the finite element mechanical model of the digital twin according to the building structural mechanics analysis standard. A full-model mechanical response simulation calculation is performed on the digital twin after applying extreme loads to obtain the maximum stress, maximum displacement, and internal force redistribution mechanical parameters of each key structural component under extreme loads. Then, the corresponding safety index is calculated, and the safety level of each component under extreme loads is determined. Based on simulation calculation results, a forward-looking risk assessment is conducted to analyze high-risk components, weak points and possible failure modes of building structures under extreme loads, and a risk assessment report is generated. The report includes working condition parameters, simulation results, a list of high-risk components, risk level assessment and preventive measures recommendations.

10. A system for constructing and evolving a digital twin of the safety status of existing building structures, characterized in that, The method for constructing and evolving a digital twin of the safety status of existing building structures as described in any one of claims 1-9 includes: The model coding and monitoring task linking module is used to code each key structural component in the engineering BIM model of an existing building according to a predefined BIM model component unique coding standard; based on the structural safety risk assessment results, it compiles monitoring task items that include monitored physical quantities, sensor information, and early warning thresholds; through component coding, it links the monitoring task items with the corresponding components in the BIM model to form an initial digital twin framework with identification and monitoring tasks; at the same time, based on the geometric and material information of the BIM model, it generates a parametric finite element reference mechanical model, which serves as the mechanical core of the digital twin; The monitoring data acquisition and transmission module is used to periodically acquire multi-dimensional time-series monitoring data reflecting the structural status through a sensor network deployed at key structural parts of existing buildings and bound to monitoring tasks, and transmit the monitoring data to the central processing server in real time through a wireless communication protocol. The data assimilation and digital twin dynamic evolution calculation module is used to establish a data assimilation algorithm engine. It uses the finite element reference mechanical model and its parameters initialized in step S1 as the state vector and the real-time monitoring data obtained in step S2 as the observation value. It performs iterative calculations through the unscented Kalman filter algorithm to obtain the posterior optimal estimate of the state vector, thereby dynamically correcting the physical state parameters (material properties, boundary conditions, internal force distribution, etc.) of the corresponding components in the digital twin and driving the finite element reference mechanical model to recalculate, realizing the synchronous evolution of the mechanical state of the digital twin and the solid structure. The safety assessment and visualization early warning module is used to calculate the real-time safety indicators of each component based on the updated digital twin status in step S3; compare the real-time safety indicators with preset multi-level safety thresholds, and determine the safety level of each component according to the numerical range; based on the safety level determination results, classify and color-code the corresponding components on the lightweight 3D BIM model, and display early warning information in real time; and use the updated digital twin to apply simulated future extreme load conditions to predict the structural response, thereby achieving a forward-looking risk assessment of the building structure.

Images