Digital twinborn autonomous optimization system based on building structure quality monitoring

Abstract

The invention discloses a digital twin autonomous optimization system based on building structure quality monitoring. According to the system, structure state data, environmental influence data, material performance data and historical record data are comprehensively collected through a multi-source data collection module; feature data such as a structure degradation rate and a material aging coefficient are extracted through an edge calculation preprocessing module; a digital twinborn body is constructed through a multi-layer fusion module, geometric layer fusion achieves crack grading labeling, physical layer fusion corrects parameters such as material elasticity modulus, and behavior layer fusion updates the structural damping ratio to reproduce dynamic behaviors. Outputting a structure health index, residual life prediction and maintenance suggestion through an autonomous optimization decision module; and in combination with a dynamic iteration updating module, the system is ensured to continuously adapt to the structure change through validity evaluation and parameter adjustment. The accuracy of building structure quality evaluation and the scientificity of maintenance decision making are improved, and the method is suitable for full-life-cycle management of various buildings.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of digital twinning, and relates to a digital twinning autonomous optimization system based on building structure quality monitoring. BACKGROUND

[0002] With the accelerated digital transformation of the construction industry, the importance of structure life cycle quality control is increasingly prominent. Current building structure quality monitoring still faces many challenges, among which structure defect identification lag, poor environmental adaptability and lack of full life cycle dynamic adaptation are particularly prominent, which has become a key bottleneck restricting the development of intelligent construction.

[0003] Existing building structure quality monitoring methods can meet basic needs, but have obvious limitations: on the one hand, traditional methods rely on manual inspection and single-point sensors, data collection is fragmented, it is difficult to achieve global real-time monitoring, and the identification accuracy of hidden defects such as cracks and steel corrosion is insufficient, which easily leads to missed judgment of safety hazards; on the other hand, existing monitoring models are mostly static, cannot effectively integrate environmental data such as temperature, humidity and load, and the evaluation accuracy in complex environments is greatly reduced, and lack of dynamic iteration mechanism, making it difficult to adapt to material aging and structure changes in the building life cycle.

[0004] In view of this, in order to solve the problems raised in the background art, a building structure quality autonomous optimization method based on digital twinning is proposed. SUMMARY

[0005] The purpose of the application can be achieved by the following technical solutions: the application provides a digital twinning autonomous optimization system based on building structure quality monitoring, which comprises a multi-source data acquisition module, an edge computing preprocessing module, a multi-level fusion module, an autonomous optimization decision module and a dynamic iterative updating module.

[0006] The multi-source data acquisition module acquires multi-element data of the target building structure by marking the specified building as the target building, including structure state data, environmental influence data, material performance data and historical record data; The edge computing preprocessing module extracts feature data by performing edge computing preprocessing on the multi-element data; The multi-level fusion module constructs a digital twinning body that is synchronously mapped with the physical building by multi-level fusion of the feature data; The autonomous optimization decision module makes autonomous optimization decisions based on the constructed digital twinning body, and generates quality level judgments and maintenance suggestions; The dynamic iterative updating module performs dynamic iterative updating on the digital twinning body and autonomous optimization decisions based on monitoring feedback.

[0007] Preferably, the acquisition of the structural state data includes: real-time acquisition of stress values ​​at each monitoring point using a fiber optic stress sensor; and extraction of crack length, width, and area using a high-definition industrial camera combined with image processing algorithms. The total area S of the region where the crack is located is determined by a laser rangefinder. Preferably, the collection of the environmental impact data includes: collecting ambient temperature and relative humidity through temperature and humidity sensors; collecting the concentration of corrosive components in the air through a chloride ion sensor; collecting real-time load through a pressure sensor; and collecting ultraviolet intensity through an ultraviolet sensor. Preferably, the acquisition of the material performance data includes: obtaining the ratio of the corroded area of ​​the reinforcing bars to the total surface area using an electromagnetic induction type reinforcing bar corrosion detector; obtaining the surface hardness of the concrete using a rebound hammer, and converting it into the measured strength of the concrete. By retrieving the concrete design strength from the design drawings The connection preload was tested using a torque wrench; based on historical test data, the aging rate was obtained by fitting the curves of concrete carbonation depth and steel corrosion thickness over time. A structural health index (HI) is defined using crack parameters and stress values, specifically as follows: in, The average stress value. The yield stress of the material. Where is the crack area, and S is the total area of ​​the monitored area.

[0008] Preferably, the step of the edge computing preprocessing module extracting feature data includes: The multi-source data is cleaned, spatiotemporally aligned, and format-converted to obtain preprocessed data; Based on the preprocessed data, the structural degradation rate was calculated to be: ; in, The structural health index at the initial moment. The current health index, This refers to the interval time. The material aging coefficient is ; in, For the measured strength of concrete, The design strength of the concrete.

[0009] Preferably, the step of multi-level fusion to construct a digital twin that is synchronously mapped to the physical building includes: S1. Perform geometric layer fusion, fuse the 3D building model with the structural defect data in the structural state data to generate a geometric twin with defect features, classify the defects into micro cracks, medium cracks and severe cracks by using dual thresholds of crack length and crack width, and integrate the graded and labeled crack data with the original 3D model to form a geometric twin that fully reflects the geometric shape and defect distribution of the physical building. The threshold for the microcracks is: crack length ≤ 50 mm and crack width ≤ 0.2 mm; The threshold for a moderate crack is: 50mm < crack length ≤ 200mm and 0.2mm < crack width ≤ 0.5mm; The threshold for severe cracks is: crack length > 200 mm or crack width > 0.5 mm; S2. Perform physical layer fusion, combining the material performance data with the geometric twin to construct a physical twin with physical properties. Specifically, based on the "geometric twin with defect features" generated by geometric layer fusion, material performance data is fused to give the geometric model physical properties, and a physical twin with physical properties is constructed. The material's physical parameters are corrected using on-site measured data, with the correction formula for the material's elastic modulus E being: in, The design elastic modulus of the material, The aging factor is... S3. Perform behavioral layer fusion, which integrates the structural dynamic response data in the structural state data with the environmental impact data to construct a behavioral twin that can reproduce the dynamic behavior of the structure. Specifically, based on the "physical twin with physical attributes" generated by physical layer fusion, the structural dynamic response data in the structural state data and the environmental impact data are integrated to construct a behavioral twin that can reproduce the dynamic behavior of the structure, thereby realizing virtual simulation of the dynamic response of the physical building under different environmental conditions. The structural damping ratio of the behavioral twin The following formula is used for real-time updates: in, The initial damping ratio of the structure is a fixed reference value. This is the temperature influence coefficient. This is the frequency influence coefficient. Let be the load influence factor, where , , , This represents the change in ambient temperature relative to the initial temperature. Where F is the natural frequency of the structure and F is the magnitude of the real-time load.

[0010] S4. Integrate the geometric twin, physical twin, and behavioral twin to construct the digital twin; Preferably, the autonomous optimization decision-making module is based on the quality level judgment and maintenance suggestions generated by the digital twin, specifically as follows: Based on the building type, load the adapted intelligent optimization model onto the digital twin; The parameters of the digital twin are input into the intelligent optimization model for inference, and the structural health index HI and remaining life prediction are output. Remaining lifetime prediction is based on crack propagation rate calculation, specifically: in, This is the initial crack length. The critical crack length is taken as a fraction of the component's cross-sectional thickness. Material constants The shape factor Y = 1.12 and the fatigue index m = 3.2. The stress intensity factor amplitude; Based on the structural health index, the quality level of the building structure is determined according to a preset threshold. HI≥0.8: Classified as Level 1; 0.5≤HI<0.8: Classified as Level 2; HI<0.5: Classified as Level 3; Generate targeted maintenance recommendations based on the quality level of the warning or danger status; Calculate the optimization benefit value, which is a quantitative value of the overall maintenance cost and structural safety improvement, specifically: in To maintain the difference in health index before and after, B>0 indicates that the maintenance plan is feasible.

[0011] Preferably, the dynamic iterative update step includes: The effectiveness coefficient E is calculated and its effect is evaluated as follows:

[0012] When E ≥ 0.2: it is classified as Level 1; When 0.1 ≤ E < 0.2: it is classified as Level 2; When E < 0.1: it is classified as Level 3; When the building structure changes, the geometric, physical, and behavioral parameters of the digital twin are automatically adjusted to maintain synchronization with the physical building.

[0013] Preferably, the digital twin is constructed by integrating the geometric twin, physical twin, and behavioral twin. Among them, the synchronization mapping accuracy of the digital twin Calculated once a day: Where M is the total number of parameters involved in the verification. For twin parameters, For entity parameters.

[0014] Preferably, a computer device includes a memory and a processor, and an algorithm iteration step size. The convergence condition is met, specifically: and in, For model parameters, The objective function is to ensure stable convergence of the iterative process.

[0015] The technical effects and advantages of this invention are as follows: 1. Improve the accuracy of quality assessment: By collecting multi-source data and mapping digital twins with high precision, combined with quantitative indicators such as structural health index and crack classification, an objective assessment of the building structure status can be achieved, solving the problems of strong subjectivity and insufficient accuracy of traditional manual inspection. The mapping accuracy can reach more than 90%.

[0016] 2. Optimize the scientific nature of maintenance decisions: Based on the dynamic simulation and remaining life prediction of digital twins, combined with the effectiveness evaluation of maintenance recommendations, targeted solutions can be generated for different quality levels (good, warning, dangerous), avoiding over-maintenance or under-maintenance, and reducing maintenance costs by 15%-20%.

[0017] 3. Achieve dynamic adaptation throughout the entire lifecycle: By dynamically iterating and updating modules, the system integrates data such as structural changes and environmental changes in real time, continuously optimizes model parameters and decision-making logic, and ensures that the system adapts to the performance changes throughout the entire lifecycle of the building, providing continuous and reliable technical support for smart construction and operation and maintenance. Attached Figure Description

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

[0019] Figure 1 This is a schematic diagram of the module connection of the present invention.

[0020] Figure 2 This is a schematic diagram illustrating the process of implementing the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of the present invention.

[0022] The specific process is based on the flowchart, as shown below. Figure 2 As shown, it includes the following steps: The multi-source data acquisition module collects multi-dimensional data of the target building structure by marking a specified building as the target building, including structural status data, environmental impact data, material performance data, and historical data. In this embodiment, structural status data reflects the current mechanical performance and defect distribution of the building. Data is collected through a distributed stress-strain monitoring unit. Fiber grating stress sensors are deployed at key stress-bearing locations such as beam mid-span, column base, and joints of the target building. The sampling frequency is set to 1Hz to collect stress values ​​(unit: MPa) at each monitoring point in real time. The average stress value of each monitoring area is automatically calculated daily. As a basic parameter for structural health assessment, a high-definition industrial camera (resolution ≥ 5 million pixels) is used to scan the building surface, and image processing algorithms are used to extract the crack length, width, and area. Meanwhile, the total area S of the region where the crack is located is determined by a laser rangefinder, providing data for calculating the structural health index.

[0023] According to the formula: in, The yield stress of building materials (such as C30 concrete) Q355 steel (This information needs to be entered into the system in advance based on the type of building structure materials.) The average stress value of the monitoring area Let S be the total area of ​​all cracks in the monitored area. For example, the monitored area of ​​a concrete beam... , , If S = 100000 mm², then: .

[0024] Furthermore, environmental factors directly affect the durability and mechanical performance of building structures, requiring real-time monitoring through environmental impact sensing units. Temperature and humidity sensors should be installed on the building facade and key indoor areas (such as basements and roofs) to collect ambient temperature (T, unit: °C) and relative humidity (RH, unit: %) at a sampling frequency of 0.5 Hz, used to analyze the impact of temperature changes on structural deformation. Chloride ion sensors should be deployed on concrete surfaces to collect the concentration of corrosive components in the air (C, unit: mg / m³) at a sampling frequency of once per hour, assessing the risk of material corrosion. Pressure sensors should be installed on load-bearing components such as floor slabs and roofs to collect real-time loads (F, unit: kN / m²) at a sampling frequency of 1 Hz, focusing on monitoring load changes under conditions of high population density and equipment stacking. Ultraviolet sensors should be installed on the building roof to collect ultraviolet radiation intensity (UV, unit: W / m²), recording the cumulative radiation daily to assess the aging rate of exterior wall materials.

[0025] Furthermore, material performance data reflects the inherent quality of the building structure. This data is collected periodically by a material performance sensing unit. An electromagnetic induction rebar corrosion detector is used to measure the ratio of the corroded area of ​​the rebar to the total surface area on beam and column members, conducted quarterly to assess the degree of degradation in the rebar's mechanical properties. Concrete surface hardness is measured on-site using a rebound hammer, and the actual concrete strength is calculated from this data. Simultaneously retrieve the concrete design strength from the design drawings. As the basis for calculating the material aging coefficient, a torque wrench is used to test the connection preload (unit: N·m) for key parts such as steel structure bolt connections and concrete joints, and the test is conducted every six months to ensure the reliability of joint force transmission. Based on historical test data, the curves of concrete carbonation depth and steel corrosion thickness over time are fitted to obtain the aging rate (k, unit: mm / year), which is used to predict the long-term trend of material performance changes.

[0026] Furthermore, historical data provides a longitudinal comparison basis for structural performance analysis. By integrating and storing historical data through a terminal, including concrete pouring strength, steel reinforcement ratio, curing days, construction temperature, etc., extracted from construction logs, it is used to trace the initial quality status of the structure; record past crack repair time, reinforcement material type (such as carbon fiber cloth, epoxy resin), replacement component model, etc., as a reference for current maintenance plan; collect structural damage assessment reports after disasters such as earthquakes, typhoons, and rainstorms, including data such as crack development and component deformation, to verify the disaster response simulation accuracy of the digital twin.

[0027] The edge computing preprocessing module extracts feature data by performing edge computing preprocessing on multi-dimensional data; Data preprocessing steps include: data cleaning, spatiotemporal alignment, and format conversion; It should be noted that the data cleaning specifically refers to: based on Criteria for excluding stress values ​​exceeding [specific value] ( Abnormal samples with a crack width >5mm (indicating sensor malfunction) are identified as historical data standard deviation. For data such as temperature, humidity, and load, values ​​exceeding the physically reasonable range (e.g., temperature >60℃ or <-20℃) are removed. When the data missing rate is <5%, the average of adjacent time points is used to fill the gap. When the missing rate is ≥5%, the time period is marked as invalid and data needs to be collected again.

[0028] It should be noted that the spatiotemporal alignment specifically refers to spatial alignment and temporal alignment. Spatial alignment involves uniformly converting all sensor data to the building's global coordinate system (with the center point of the foundation as the origin, the X-axis representing the building's length, the Y-axis representing the width, and the Z-axis representing the height), and eliminating sensor installation position deviations through a coordinate transformation matrix. Temporal alignment uses the stress sensor timestamp as a reference and synchronizes data at different frequencies, such as temperature and humidity, load temperature and humidity, and load, through linear interpolation to ensure that the timestamp deviation of all data is ≤0.1s.

[0029] It should be noted that the format conversion involves converting image data (such as crack images) into a pixel coordinate matrix with a uniform resolution of 1024×1024; and converting sensor data into JSON format, which includes fields such as parameter name, value, timestamp, and acquisition location, to facilitate subsequent fusion calculations.

[0030] Furthermore, feature data extraction, based on preprocessed data, calculates characteristic parameters reflecting the structural state and material properties, providing input structural degradation rate and material aging coefficient for digital twin construction. Specifically: The structural degradation rate is ;in, The structural health index at the initial moment. The current health index, This refers to the interval time. For example, when the initial time of a building is 0, 2 years ,but: ; This further indicates that the structure's health status declines by an average of 0.1% per year.

[0031] The material aging coefficient is ;in, For the measured strength of concrete, This refers to the design strength of the concrete. For example, the design strength of C30 concrete. Actual strength ,but: ; Furthermore, this indicates that the material strength has decreased by 10%.

[0032] The multi-level fusion module constructs a digital twin that is synchronously mapped to the physical building by fusing feature data at multiple levels. The steps of constructing the digital twin synchronously mapped to the physical building through multi-level fusion include: S1. Perform geometric layer fusion, which merges the 3D building model with the structural defect data in the structural state data to generate a geometric twin with defect features; S2. Perform physical layer fusion, combining the material performance data with the geometric twin to construct a physical twin with physical properties; S3 performs behavioral layer fusion, fusing the structural dynamic response data in the structural state data with the environmental impact data to construct a behavioral twin capable of reproducing the structural dynamic behavior; S4. Integrate the geometric twin, physical twin, and behavioral twin to construct the digital twin; It should be noted that geometric layer fusion combines the 3D building model with structural defect data to generate a geometric twin with defect features, intuitively presenting the structural appearance and defect distribution. Specifically: Basic model construction: Based on the building BIM model as the basic framework, it includes information such as the geometric dimensions and spatial positions of components such as beams, columns, and slabs, with accuracy controlled within ±5mm; Defects are classified according to dual thresholds of crack length and width, with specific standards as follows: Microcracks: Cracks with a length ≤ 50 mm and a width ≤ 0.2 mm, mainly harmless cracks caused by material shrinkage; Medium-sized cracks: 50mm < crack length ≤ 200mm and 0.2mm < crack width ≤ 0.5mm, their development trend needs to be monitored; Severe cracks: Cracks with a length > 200 mm or a width > 0.5 mm may affect the structural bearing capacity and require immediate treatment.

[0033] Furthermore, using coordinate matching technology, the graded crack data (location, length, width, and grade) is mapped to the corresponding locations in the BIM model. Different colors are used to indicate crack grades (blue for micro-cracks, yellow for medium-sized cracks, and red for severe cracks), generating a geometric twin. For example, if a crack with a length of 150mm and a width of 0.3mm exists on the surface of a frame column, it will be marked with a yellow line at the corresponding location on the column in the geometric twin, along with the crack parameter information.

[0034] It should be noted that physical layer fusion combines material property data with the geometric twin, endowing the virtual model with physical properties, giving it mechanical properties consistent with the solid structure. Specifically, based on material property data, the physical parameters of each component in the geometric twin are corrected. The core parameter includes the elastic modulus, specifically: in, The design elastic modulus of the material (e.g., for C30 concrete, E_design = 3.0 × 10⁻⁶). 4 MPa, Q355 steel E_design = 2.06 × 10 5 MPa); The material aging coefficient; This is a material type coefficient (1.0 for concrete, 1.1 for steel structures, and 0.9 for composite materials). For example, the coefficient for a concrete beam... , , Then, after correction: Furthermore, Poisson's ratio It is set according to the material type, with 0.2 for concrete and 0.3 for steel, and is reduced by 5% every 5 years as the material ages.

[0035] Ultimate strength: Based on the steel corrosion rate and the measured strength of concrete, the tensile and compressive ultimate strengths of the component are corrected. For example, for every 10% increase in steel corrosion rate, the tensile strength is reduced by 8%.

[0036] Furthermore, the corrected physical parameters are bound to the corresponding components of the geometric twin. For example, the elastic modulus and ultimate strength are entered in the beam component properties, and the connection stiffness is entered in the node properties, so that the virtual model can truly reflect the stress characteristics of the physical structure.

[0037] It should be noted that behavioral layer fusion combines structural dynamic response data with environmental impact data to construct a behavioral twin capable of reproducing structural dynamic behavior, simulating structural responses under different environments. Specifically: The structural dynamic characteristic parameters are updated in real time based on environmental data. The core parameter is the structural damping ratio, and the calculation formula is as follows: in, , , , , This represents the change in ambient temperature relative to the initial temperature. The natural frequency of the structure, F represents the change in ambient temperature relative to the initial temperature, and F represents the real-time load magnitude.

[0038] A building has an initial temperature of 25℃, a current temperature of 35℃ (ΔT=10℃), a natural frequency of f=1.5Hz, and a real-time load of F=3kN / m². Then... That is, the damping ratio increases from the initial 0.02 to 0.028, reflecting the change in the structure's energy dissipation capacity under increased temperature and load.

[0039] Furthermore, the updated damping ratio, elastic modulus, and other parameters are input into the structural dynamics equations to simulate the dynamic response of the structure under vibration, temperature deformation, and load (such as changes in displacement amplitude and vibration frequency). For example, under strong wind loads, the vibration displacement of the roof is simulated using a behavioral twin, and the simulation accuracy is verified by comparing it with the actual displacement collected by sensors.

[0040] It should be noted that digital twin integration integrates geometric, physical, and behavioral twins to form a complete digital twin, achieving full-dimensional mapping of physical buildings. Specifically, this involves data association, model integration, and accuracy verification.

[0041] Furthermore, the data association is based on the building's global coordinate system, aligning the spatial coordinates and timestamps of the three types of twins to ensure a one-to-one correspondence between the location of geometric defects, physical parameters, and dynamic responses in time and space. For example, the location of a crack marked in the geometric twin needs to be associated with the decrease in the elastic modulus of that area in the physical twin and the stress concentration phenomenon at that location in the behavioral twin.

[0042] Furthermore, the model integration is achieved through a digital twin fusion platform, which integrates the model data, parameter data, and dynamic response data of the three types of twins into a unified database, supporting 3D visualization and parameter query. Users can click on any component in the virtual interface to view its geometric shape (including cracks), physical parameters (such as elastic modulus), and dynamic response (such as real-time stress).

[0043] Furthermore, the accuracy verification involves calculating the synchronization mapping accuracy of the digital twin daily to ensure consistency between the virtual model and the physical building. Specifically: Where M represents the total number of parameters involved in the verification (such as stress, displacement, frequency, etc.). When the accuracy is ≥0.8, the digital twin is considered valid; otherwise, the sensor data and model parameters need to be recalibrated until the accuracy requirements are met.

[0044] The autonomous optimization decision-making module makes autonomous optimization decisions based on the constructed digital twin, generating quality level judgments and maintenance suggestions. It should be noted that the autonomous optimization decision-making is based on digital twins. It uses intelligent model reasoning to determine the structural quality level and generate maintenance suggestions, providing precise guidance for building operation and maintenance. This includes: intelligent optimization model loading, model reasoning and parameter output, and quality level determination and maintenance suggestions.

[0045] Furthermore, the intelligent optimization model loading involves selecting an appropriate intelligent optimization model based on the building type, specifically divided into: Super high-rise buildings: Load the "seismic response optimization model" to focus on analyzing the dynamic response of the structure under seismic loads; Long-span bridges: Loading a "fatigue damage optimization model" to analyze component fatigue life based on vehicle load; Industrial plants: Load the "corrosion environment optimization model" and combine it with the concentration of corrosive components to predict the corrosion rate of materials.

[0046] Furthermore, the model inference and parameter output involves inputting the feature parameters of the digital twin into the intelligent optimization model and outputting key evaluation indicators, specifically: Structural Health Index (HI): Calculated according to the formula above, the range is [0,1], and the higher the value, the better the structural condition.

[0047] Remaining lifetime prediction (RUL): calculated based on crack propagation rate, specifically: in, The initial crack length (in meters). Critical crack length (unit: m, taken as 1 / 3 of the component's cross-sectional thickness); material constants The shape factor Y = 1.12 and the fatigue index m = 3.2. Stress intensity factor amplitude.

[0048] For example, an initial crack in a beam member Critical crack , ,but: This indicates that it would take approximately 83 years for the crack to expand to a critical state.

[0049] Optimization Benefit Value (B): A quantitative value reflecting the overall improvement in maintenance costs and structural safety, specifically: in To maintain the difference in health index before and after, B>0 indicates that the maintenance plan is feasible.

[0050] Furthermore, the quality level determination and maintenance recommendations specifically include: determining the quality level based on a preset threshold of the structural health index HI, including: HI≥0.8: Classified as Level 1, indicating good performance, stable structural performance, no special maintenance required, only routine inspection; 0.5≤HI<0.8: Classified as Level II, indicating a warning, indicating localized structural defects requiring targeted reinforcement; HI<0.5: Classified as Level III, indicating danger, insufficient structural bearing capacity, requiring immediate shutdown and the development of a major overhaul plan.

[0051] Based on the assessment results, maintenance recommendations are generated as follows: For the "warning" state: if the crack is a micro-crack, it is recommended to use epoxy resin grouting for repair; if the steel reinforcement corrosion rate is >10%, it is recommended to apply a rust inhibitor; if the preload of the joint is insufficient, it is recommended to retighten the bolts; For the "dangerous" state: if the stress at the mid-span of the beam exceeds the standard, it is recommended to attach carbon fiber cloth for reinforcement; if the crack at the bottom of the column is severe, it is recommended to increase the cross-section for reinforcement; if the foundation settlement exceeds the standard, it is recommended to use anchor static pressure piles for correction.

[0052] The dynamic iterative update module dynamically updates the digital twin and the autonomous optimization decision through monitoring and feedback. It should be noted that the dynamic iterative update continuously optimizes the digital twin and decision model through monitoring and feedback, ensuring that the system adapts to changes in structure and environment in the long term. Specifically, this includes: evaluating the effectiveness of maintenance suggestions, adjusting the model when the structure changes, and optimizing long-term data.

[0053] Furthermore, the effectiveness assessment of the maintenance recommendations involves calculating the effectiveness coefficient E and evaluating its effect after the maintenance plan is implemented, specifically as follows: When E≥0.2: it is judged as Level 1, indicating that it is significantly effective, the maintenance plan is reasonable, and the current strategy should be maintained; When 0.1≤E<0.2: it is judged as level two, indicating that it is partially effective and local parameters need to be optimized (such as adjusting the amount of reinforcement material). When E < 0.1: it is judged as level three, indicating invalidity, and the maintenance plan needs to be redesigned (such as changing the reinforcement method).

[0054] For example, if a building's HI was 0.6 before maintenance and HI was 0.75 after maintenance, then: If deemed significantly effective, this type of maintenance solution will continue to be used.

[0055] It should be noted that when structural changes occur to a building (such as adding floors, removing components, or replacing materials), the digital twin is automatically updated. Specifically: the geometric twin needs to modify the dimensions and positions of the changed components in the BIM model, delete the removed parts, and add new components; the physical twin needs to update the material parameters of the changed components (such as the elastic modulus and Poisson's ratio of the added steel structure); and the behavioral twin needs to recalculate the structure's natural frequency, damping ratio, and other dynamic parameters to ensure consistency with the changed physical structure.

[0056] It should be further explained that this method can periodically optimize system parameters based on long-term operating data (at least one year). Specifically, it involves adjusting the weights of structural degradation rate and material aging coefficient in the decision model using machine learning algorithms, such as increasing the weight of corrosive component concentration in a corrosive environment; retraining the intelligent optimization model with new monitoring data to update parameters such as crack propagation coefficient and fatigue index, thereby improving prediction accuracy; and ensuring that the algorithm iteration step size meets the convergence condition. and in, For model parameters, The objective function is to ensure stable convergence of the iterative process.

[0057] For example, suppose there is a key parameter in the digital twin of a building structure. (e.g., correction factor for material aging coefficient), initial value The system iteratively optimizes this parameter, aiming to minimize the prediction error of the updated model. The iterative rule is as follows: after each iteration, the new parameters... Compared with old parameters The difference must satisfy: The specific process is as follows: First iteration: old parameters The new parameters are calculated. The difference is |0.800005-0.8|=0.000005≤0.00001, which satisfies the condition, and the iteration is valid.

[0058] Second iteration: old parameters The new parameters are calculated. The difference is |0.800008-0.800005|=0.000003≤0.00001, which satisfies the condition, so continue iterating.

[0059] If a certain calculation yields Old parameters The difference is |0.80002-0.800005|=0.000015>0.00001, which does not meet the condition. The adjustment range needs to be reduced and the calculation recalculated. (For example, if adjusted to 0.800012, the difference is 0.000007, which meets the condition).

[0060] This control ensures that the parameters are updated in small increments each time, allowing the model to gradually converge to the optimal value and avoiding prediction fluctuations caused by large step sizes.

[0061] Through the coordinated operation of the aforementioned modules, this method enables autonomous optimization of building structural quality throughout its entire lifecycle. In practical applications, the mapping accuracy between the digital twin and the physical building can reach over 90%, the accuracy of structural health assessment is improved by 25%, and maintenance costs are reduced by 15% to 20%. For example, a high-rise residential building used this system to provide early warning of the risk of crack propagation in the basement beams, allowing for timely grouting reinforcement and preventing the cracks from further developing into structural safety hazards.

[0062] This implementation plan, through a closed-loop design involving multi-source data acquisition, edge computing preprocessing, multi-level fusion, autonomous optimization decision-making, and dynamic iterative updates, comprehensively integrates all parameters and formulas in the claims, achieving intelligent and precise control over building structural quality and providing effective technical support for smart construction and operation and maintenance.

[0063] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other. In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A digital twin autonomous optimization system based on building structure quality monitoring, characterized in that, include: The multi-source data acquisition module collects multi-dimensional data of the target building structure by marking a specified building as the target building, including structural status data, environmental impact data, material performance data, and historical data. The edge computing preprocessing module extracts feature data by performing edge computing preprocessing on multi-dimensional data; The multi-level fusion module constructs a digital twin that is synchronously mapped to the physical building by fusing feature data at multiple levels. The steps of the multi-layered fusion to construct a digital twin that is synchronously mapped to the physical building include: S1. Perform geometric layer fusion, fuse the 3D building model with the structural defect data in the structural state data to generate a geometric twin with defect features, classify the defects into micro cracks, medium cracks and severe cracks by using dual thresholds of crack length and crack width, and integrate the graded and labeled crack data with the original 3D model to form a geometric twin that fully reflects the geometric shape and defect distribution of the physical building. The threshold for the microcracks is: crack length ≤ 50 mm and crack width ≤ 0.2 mm; The threshold for a moderate crack is: 50mm < crack length ≤ 200mm and 0.2mm < crack width ≤ 0.5mm; The threshold for severe cracks is: crack length > 200 mm or crack width > 0.5 mm; S2. Perform physical layer fusion, combining the material performance data with the geometric twin to construct a physical twin with physical properties. Specifically, based on the "geometric twin with defect features" generated by geometric layer fusion, material performance data is fused to give the geometric model physical properties and construct a physical twin with physical properties. The material's physical parameters are corrected using on-site measured data, with the correction formula for the material's elastic modulus E being: ； in, The design elastic modulus of the material, The aging factor is... S3. Perform behavioral layer fusion, which integrates the structural dynamic response data in the structural state data with the environmental impact data to construct a behavioral twin that can reproduce the dynamic behavior of the structure. Specifically, based on the "physical twin with physical attributes" generated by physical layer fusion, the structural dynamic response data in the structural state data and the environmental impact data are integrated to construct a behavioral twin that can reproduce the dynamic behavior of the structure, thereby realizing virtual simulation of the dynamic response of the physical building under different environmental conditions. The structural damping ratio of the behavioral twin The following formula is used for real-time updates: ； in, The initial damping ratio of the structure is a fixed reference value. This is the temperature influence coefficient. This is the frequency influence coefficient. Let be the load influence factor, where , , , This represents the change in ambient temperature relative to the initial temperature. Where F is the natural frequency of the structure, and F is the magnitude of the real-time load. S4. Integrate the geometric twin, physical twin, and behavioral twin to construct the digital twin; The autonomous optimization decision-making module makes autonomous optimization decisions based on the constructed digital twin, generating quality level judgments and maintenance suggestions. The autonomous optimization decision-making module is based on the quality level judgment and maintenance suggestions generated by the digital twin, specifically as follows: Based on the building type, load the adapted intelligent optimization model onto the digital twin; The parameters of the digital twin are input into the intelligent optimization model for inference, and the structural health index HI and remaining life prediction are output. Remaining lifetime prediction is based on crack propagation rate calculation, specifically: ； in, The initial crack length is... The critical crack length is taken as a fraction of the component's cross-sectional thickness. Material constants The shape factor Y = 1.12 and the fatigue index m = 3.

2. The stress intensity factor amplitude; Based on the structural health index, the quality level of the building structure is determined according to a preset threshold. HI≥0.8: Classified as Level 1; 0.5≤HI<0.8: Classified as Level 2; HI<0.5: Classified as Level 3; Generate targeted maintenance recommendations based on the quality level of the warning or danger status; Calculate the optimization benefit value, which is a quantitative value of the overall maintenance cost and structural safety improvement, specifically: ； in To maintain the difference in health index before and after, B>0 indicates that the maintenance plan is feasible; The dynamic iterative update module dynamically updates the digital twin and the autonomous optimization decision through monitoring and feedback.

2. The digital twin autonomous optimization system based on building structure quality monitoring according to claim 1, characterized in that, The acquisition of the structural state data includes: real-time acquisition of stress values ​​at each monitoring point using a fiber optic stress sensor; and extraction of crack length, width, and area using a high-definition industrial camera combined with image processing algorithms. The total area S of the region where the crack is located is determined by a laser rangefinder. The environmental impact data is collected, including: ambient temperature and relative humidity collected by temperature and humidity sensors; concentration of corrosive components in the air collected by chloride ion sensors; real-time load collected by pressure sensors; and ultraviolet intensity collected by ultraviolet sensors. The material performance data is collected, including: obtaining the ratio of the corroded area of ​​the reinforcing bars to the total surface area using an electromagnetic induction type reinforcing bar corrosion detector; obtaining the surface hardness of the concrete using a rebound hammer, and converting it into the measured strength of the concrete. By retrieving the concrete design strength from the design drawings The connection preload was tested using a torque wrench; based on historical test data, the aging rate was obtained by fitting the curves of concrete carbonation depth and steel corrosion thickness over time. A structural health index (HI) is defined using crack parameters and stress values, specifically as follows: ； in, The average stress value. The yield stress of the material. Where is the crack area, and S is the total area of ​​the monitored area.

3. The digital twin autonomous optimization system based on building structure quality monitoring according to claim 1, characterized in that, The steps for extracting feature data by the edge computing preprocessing module include: The multi-source data is cleaned, spatiotemporally aligned, and format-converted to obtain preprocessed data; Based on the preprocessed data, the following calculations were performed: The structural degradation rate is ; in, The structural health index at the initial moment. The current health index, This refers to the interval time. The material aging coefficient is ; in, For the measured strength of concrete, The design strength of the concrete.

4. The digital twin autonomous optimization system based on building structure quality monitoring according to claim 1, characterized in that, The steps of the dynamic iterative update include: The effectiveness coefficient E is calculated and its effect is evaluated as follows: ； When E ≥ 0.2: it is classified as Level 1; When 0.1 ≤ E < 0.2: it is classified as Level 2; When E < 0.1: it is classified as Level 3; When the building structure changes, the geometric, physical, and behavioral parameters of the digital twin are automatically adjusted to maintain synchronization with the physical building.

5. The digital twin autonomous optimization system based on building structure quality monitoring according to claim 1, characterized in that, The digital twin is specifically constructed by integrating the geometric twin, physical twin, and behavioral twin. Among them, the synchronization mapping accuracy of the digital twin Calculated once a day: ； Where M is the total number of parameters involved in the verification. For twin parameters, For entity parameters.

6. A computer device, characterized in that: Including memory and processor, algorithm iteration step size The convergence condition is met, specifically: and ; in, For model parameters, The objective function is to ensure stable convergence of the iterative process.

Images