Village and town building seismic resilience dynamic monitoring method and system based on digital twinning

Abstract

The application discloses a village and town building seismic resilience dynamic monitoring method and system based on digital twinning, and relates to the technical field of building seismic resilience evaluation. The method comprises the following steps: acquiring historical building structure data of a target building, and constructing a prior structure feature database; dividing construction quality parameters into a set of calibratable parameters and a set of uncalibratable parameters, and constructing a state quantile migration matrix; acquiring measured calibration parameter values of the calibratable parameters and mapping them into quantile states; index querying the state quantile migration matrix, acquiring conditional probability distribution of the uncalibratable parameters, and defining predicted calibration parameter values; combining the measured and predicted parameter values to construct a digital twin building, simulating structural response under seismic working conditions, and evaluating and acquiring seismic resilience monitoring results. The application solves the technical problems in the prior art that hidden structure parameters of village and town self-built brick-concrete structures cannot be reliably calibrated, and that resilience evaluation does not consider post-earthquake repair resource constraints of villages and towns.

Description

Technical Field

[0001] This invention relates to the field of building seismic toughness assessment technology, specifically to a method and system for dynamic monitoring of the seismic toughness of rural buildings based on digital twins. Background Technology

[0002] In rural areas, there are numerous two- to three-story self-built brick-concrete structures. These buildings generally lack formal design drawings, and the construction process is carried out without third-party supervision. The construction quality mainly depends on the experience and habits of local craftsmen, and the proportions of building materials, especially mortar, are highly arbitrary and regionally specific. In digital twin modeling, key structural parameters such as masonry mortar strength, masonry method, actual spacing of stirrups in structural columns, and actual installation of tie bars between walls and structural columns cannot be directly and completely obtained through conventional non-destructive testing methods. This has become a core bottleneck for the promotion of digital twin technology in rural buildings.

[0003] Existing methods for calibrating building digital twin parameters primarily rely on Bayesian model updates based on measured response data. This method uses structural dynamic response data collected by sensors and a Bayesian inference framework to update uncertain parameters in the model. However, this method has limitations when applied to self-built brick-concrete structures in villages and towns: it requires sufficient measured sensor data for updates, making it unsuitable for village and town buildings without sensors or with insufficient sensor coverage; it suffers from parameter indiscernibility when the amount of information in the measurable response data is insufficient to distinguish the contributions of multiple parameters; and it treats each building as an independent entity, ignoring the statistical commonalities in construction quality among self-built houses in the same region and at the same time. Regarding resilience assessment, existing assessment frameworks mainly focus on structural damage and repair time, failing to incorporate the actual availability of repair resources and thus failing to reflect the true constraints of post-earthquake recovery for village and town buildings. Summary of the Invention

[0004] This invention provides a method and system for dynamic monitoring of the seismic toughness of rural buildings based on digital twins. It is used to solve the technical problems in the prior art where the hidden structural parameters of the digital twin model of self-built brick-concrete structures in rural areas cannot be reliably calibrated due to the lack of complete design data and measured data, and where the toughness assessment does not take into account the resource constraints of post-earthquake repair in rural areas.

[0005] In a first aspect, the present invention provides a method for dynamic monitoring of the seismic toughness of rural buildings based on digital twins, the method comprising:

[0006] Obtain historical building construction data for the target region corresponding to the target building, extract regional construction quality parameters accordingly, and construct a regional prior construction feature database based on probability distribution;

[0007] Based on the building construction data acquisition scheme of the target building, the construction quality parameters are divided into a set of calibrable parameters and a set of non-calibrable parameters, and a state quantile migration matrix is ​​constructed in conjunction with the historical building construction data.

[0008] The measured calibration parameter values ​​of each calibrable parameter of the target building are obtained, and the feature database is constructed based on the prior of the region and mapped to the corresponding quantile state.

[0009] Using the quantile states of the calibrable parameters as indexes, query the state quantile transition matrix to obtain the conditional probability distribution of each uncalibrable parameter at each quantile state, and define the predicted calibration parameter values ​​of each uncalibrable parameter based on the conditional probability distribution.

[0010] By combining the measured calibration parameter values ​​and the predicted calibration parameter values, a digital twin building for the target building is constructed. The structural response is simulated under the set seismic loading conditions, and the seismic toughness monitoring results are obtained based on the corresponding evaluation of the structural response simulation results.

[0011] Secondly, the present invention also provides a dynamic monitoring system for the seismic toughness of rural buildings based on digital twins, the system comprising:

[0012] The regional prior database construction module is used to obtain historical building construction data of the target region corresponding to the target building, and extract regional construction quality parameters accordingly to construct a regional prior construction feature database based on probability distribution.

[0013] The parameter partitioning and migration matrix construction module is used to divide the construction quality parameters of the target building into a set of calibrable parameters and a set of non-calibrable parameters according to the building construction data acquisition scheme, and to construct a state quantile migration matrix in combination with the historical building construction data.

[0014] The quantile state mapping module is used to obtain the measured calibration parameter values ​​of each calibrable parameter of the target building, and to map the corresponding quantile state to the feature database constructed based on the prior of the region.

[0015] The uncalibrable parameter inference module is used to query the state quantile transition matrix with the quantile state of the calibrable parameter as the index, obtain the conditional probability distribution of each uncalibrable parameter in each quantile state, and define the predicted calibration parameter value of each uncalibrable parameter based on the conditional probability distribution.

[0016] The digital twin simulation and resilience evaluation module is used to combine the measured calibration parameter values ​​and the predicted calibration parameter values ​​to construct a digital twin building for the target building, simulate the structural response under the set seismic loading conditions, and obtain the seismic resilience monitoring results based on the corresponding evaluation of the structural response simulation results.

[0017] One or more technical solutions provided in this invention have at least the following technical effects or advantages:

[0018] First, this invention establishes a regional prior structural feature database of buildings of the same structural type within a target area, quantifies the regional statistical regularities of construction quality parameters in the form of probability distributions, and establishes quantile state transition relationships from calibrable parameters to uncalibrable parameters through a state quantile transition matrix. Compared with existing technologies that treat each building as an independent entity and rely on sensor measurement data for Bayesian updates, this invention utilizes the statistical commonalities in construction habits and material supply among similar buildings in the same region to achieve statistical inference of hidden structural parameters such as mortar strength, masonry method, stirrup spacing, and tie bar placement without the need for large-scale destructive testing, transforming the problem of scarce individual data into a regional statistical inference problem.

[0019] Second, this invention constructs three sub-models—a masonry wall slab mechanical model, a structural column-beam constraint effect model, and a wall structural column tie bar connection model—by using the predicted values ​​of the inferred uncalibrable parameters and the measured calibrable parameters as inputs to the digital twin model. Compared to existing technologies that update structural stiffness parameters solely based on measured response data, this invention uses the predicted calibration parameters of the uncalibrable parameters as inputs for each of the three sub-models, explicitly embedding the physical causal relationship between the parameter inference results and the mechanical model into the digital twin modeling process.

[0020] Third, this invention introduces the dimension of repair resource availability into resilience assessment. By acquiring repair resource information for the target area through big data, it quantifies the actual supply and demand ratio of building materials, skilled workers, and funds. Compared with existing technologies that only assess structural damage levels and theoretical repair time, the repair resource availability index can identify risk scenarios that are "technically repairable but resource-unrecoverable," making the resilience assessment results more reflective of the actual situation of post-earthquake recovery in rural buildings. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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.

[0022] Figure 1 is a flowchart illustrating the dynamic monitoring method for seismic resilience of rural buildings based on digital twins provided in an embodiment of the present invention.

[0023] Figure 2 is a schematic diagram of the scheme for dynamic monitoring of seismic toughness of rural buildings based on digital twins provided in an embodiment of the present invention;

[0024] Figure 3 is an example data table of regional structural construction quality parameters of the dynamic monitoring method for seismic toughness of rural buildings based on digital twins provided in an embodiment of the present invention.

[0025] Figure 4 is a schematic diagram of the structure of the dynamic monitoring method for seismic toughness of rural buildings based on digital twins provided in an embodiment of the present invention;

[0026] The diagram is labeled as follows: Module 11 for constructing the regional prior database, Module 12 for constructing the parameter partitioning and migration matrix, Module 13 for mapping the quantile state, Module 14 for inferring uncalibrated parameters, and Module 15 for digital twin simulation and resilience evaluation. Detailed Implementation

[0027] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0028] Example 1, as shown in Figure 1, provides a flowchart of a dynamic monitoring method for the seismic toughness of rural buildings based on digital twins; as shown in Figure 2, provides a scheme logic diagram of a dynamic monitoring method for the seismic toughness of rural buildings based on digital twins, the method including:

[0029] S100: Obtain historical building construction data for the target region corresponding to the target building, and extract the corresponding regional construction data.

[0030] Quality parameters are used to construct a region prior feature database based on probability distribution.

[0031] In rural areas, there are numerous two- to three-story self-built brick-concrete structures. These buildings generally lack formal design drawings and construction quality acceptance records. Hidden structural parameters such as mortar strength, masonry methods, actual spacing of stirrups in structural columns, and the placement of tie bars between walls and structural columns cannot be directly obtained through non-destructive testing. However, similar self-built houses constructed in the same region and at the same time exhibit significant statistical commonalities in construction habits, material supply, and craftsmanship. Local craftsmen have developed relatively fixed masonry habits through long-term practice, building material suppliers are relatively stable, and the dispersion of construction quality statistically follows a certain probability distribution.

[0032] Step S100 provided in this embodiment of the invention includes:

[0033] Within the target area where the target building is located, select multiple buildings of the same structural type as the target building as sample buildings;

[0034] On-site and historical data of the structural construction quality characteristics of each sample building were collected to obtain hidden structural parameter samples and corresponding directly measurable parameter samples of each building, which were used as the structural construction quality parameters of the area.

[0035] For each of the hidden structural parameter samples and the corresponding directly measurable parameter samples, a probability distribution fitting is performed to obtain a regional prior structural feature database characterized by the probability distribution.

[0036] The specific implementation method is as follows:

[0037] First, within the target area where the target building is located, select multiple buildings of the same structural type as the target building as sample buildings. The target area is an administrative division centered on the target building, such as the same township or village, or it can be divided into a larger area based on geological conditions and architectural traditions. "Same structural type" refers to buildings of the same 2-3 story brick-concrete structure as the target building and built around the same time. The number of sample buildings needs to balance the accuracy requirements of statistical inference with the feasibility of on-site data collection. Too few samples will result in insufficient statistical representativeness for probability distribution fitting, while too many samples will lead to excessively high on-site collection costs. The preset number of sample buildings is no less than 30, ensuring at least 30 independent sample values ​​for each structural construction quality parameter, thus meeting the large sample requirement for parameter probability distribution fitting.

[0038] On-site and historical data collection was conducted for the structural construction quality characteristics of each sample building. On-site data collection primarily employed micro-destructive testing and visual inspection. Micro-destructive testing involved locally chiseling open non-critical areas of each sample building. These non-critical areas were selected as those that would not affect structural safety, such as the wall under a non-load-bearing window or the top of a gable wall. After chiseling open the mortar joints, mortar fullness was tested, and mortar samples were taken for compressive strength testing to obtain the mortar compressive strength value in megapascals (MPa). After chiseling open the protective layer of the structural columns, the stirrups and longitudinal reinforcement were exposed. The stirrup spacing and rebar diameter were measured using calipers in millimeters. At the junction of the wall and the structural column, the spacing and diameter of the tie bars were checked. Visual inspection involved observing the masonry method of the exposed areas and recording the ratio of header and stretcher brick layers; observing the stirrup spacing of the exposed parts of the structural columns through weak points in the protective layer or construction openings; and measuring the building's geometric dimensions, including bay width, depth, floor height, and wall thickness. The specific method for collecting historical data is as follows: collect historical earthquake data within the target area.

[0039] Construction quality data from hazard investigation reports, dilapidated building appraisal reports, and rural housing renovation project acceptance records were used to extract historical data related to concealed structural parameters and directly measurable parameters.

[0040] After the above data collection was completed, for each sample building, a set of hidden structural parameters and a set of directly measurable parameters were obtained. The hidden structural parameters included measured values ​​of masonry method, mortar fullness, mortar compressive strength, spacing of stirrups in structural columns, diameter of stirrups in structural columns, spacing of tie bars, and diameter of tie bars. The directly measurable parameters included measured values ​​of building number of stories, story height, bay width, depth, wall thickness, and cross-sectional dimensions of structural columns. Data from all sample buildings were compiled into a regional structural construction quality parameter dataset.

[0041] Then, probability distribution fitting is performed on each sample of concealed structural parameters and its corresponding directly measurable parameter samples. The purpose of probability distribution fitting is to infer the statistical distribution pattern of each parameter among similar buildings in the target area from the sample data, and to express the probability density function of the parameter taking different values.

[0042] For parameters that are continuously variable and non-negative, a log-normal distribution is used for fitting. The log-normal distribution is a common positively skewed distribution, and its probability density function is f(x)=(1 / (x ·σ ·√(2π))) ·exp(−(lnx−μ)). 2 / (2σ 2 In the formula, x is the parameter value and x>0, μ is the location parameter, and σ is the scale parameter. μ and σ are estimated as follows: after taking the natural logarithm of the sample data, calculate the mean and standard deviation of the logarithmic sample, and use these as estimates of μ and σ, respectively. The location parameter μ determines the median level of the distribution, and the scale parameter σ determines the dispersion of the distribution. Taking mortar compressive strength as an example, the mean of the mortar compressive strength sample values ​​of 32 sample buildings in a target area after logarithmic transformation is 2.85, and the standard deviation is 0.32. Therefore, the location parameter μ is taken as 2.85, and the scale parameter σ is taken as 0.32. After fitting, the median of this distribution is exp(μ)=exp(2.85)≈17.3MPa.

[0043] For a parameter that fluctuates around a common value, a normal distribution is used for fitting. The probability density function of the normal distribution is: f(x)=(Γ(α+β) / (Γ(α)·Γ(β))) ·x^(α−1) ·(1−x)^(β−1), where x is the parameter value and 0≤x≤1, α and β are two shape parameters, and Γ(·) is the gamma function. α and β are obtained from the sample data through maximum likelihood estimation. The process of maximum likelihood estimation is as follows: construct a log-likelihood function with α and β as unknowns. The log-likelihood function is equal to the sum of the log probability densities of each sample point. Use gradient descent or Newton's iteration method to search for the combination of α and β that maximizes the log-likelihood function. The ratio of the two shape parameters, α / (α+β), determines the location of the distribution mean, and the sum of the two shape parameters determines the central tendency of the distribution. Taking mortar fullness as an example, the mortar fullness sample values ​​of 32 sample buildings in a certain target area were obtained by maximum likelihood estimation, with α=12.5 and β=2.8. The mean of the distribution is α / (α+β)= 12.5 / (12.5+2.8)≈0.817, indicating that the average mortar fullness in this area is about 81.7%.

[0044] For discrete parameters with only a few typical patterns, a discrete distribution is used for fitting. The discrete distribution is not expressed as a continuous probability density function, but rather directly uses the frequency of each typical pattern appearing in the sample as the probability value of that pattern. Taking masonry patterns as an example, in 32 sample buildings in a target area, the three-straight-one-header pattern appeared 18 times, the five-straight-one-header pattern appeared 11 times, and the all-straight pattern appeared 3 times. The three probability values ​​of the discrete distribution are: the probability of three-straight-one-header = 18 ÷ 32 = 0.5625, the probability of five-straight-one-header = 11 ÷ 32 = 0.34375, and the probability of the all-straight pattern = 3 ÷ 32 = 0.09375. The probability of the three is...

[0045] The sum is 1.

[0046] For samples of parameters that can be directly measured, the same distribution type selection principle as the aforementioned hidden construction parameters is used for fitting.

[0047] After completing the distribution fitting for each type of construction quality parameter, the distribution type, distribution parameters, sample size, and goodness-of-fit index are stored together to construct a regional prior structural feature database characterized by probability distribution. The database uses the parameter name as the primary key, and each record contains a distribution type field, a distribution parameter field, a quantile function field, and a sample statistic field. The quantile function field stores the inverse cumulative distribution function of the parameter's probability distribution, used in subsequent steps to quickly map measured parameter values ​​to quantile states. The goodness-of-fit index is used to evaluate the degree of fit between the selected distribution type and the sample data, using the Anderson-Darling statistic as the testing method. The Anderson-Darling statistic is calculated as follows: after sorting the sample data from smallest to largest, the theoretical cumulative probability value of each sorted sample point under the assumed distribution is calculated. The difference between the empirical cumulative probability and the theoretical cumulative probability of each sample point is then weighted and integrated. The weighting is larger at the tails of the distribution, making the test more sensitive to tail bias. The smaller the Anderson-Darling statistic, the higher the goodness-of-fit. When the Anderson-Darling statistic is less than the critical value for the corresponding distribution type and sample size, the distribution fit is considered acceptable.

[0048] For example, taking a certain township as the target area, 32 two-story brick-concrete self-built houses were selected as sample buildings. After performing micro-destructive testing on each sample building, 32 measured sample values ​​of the stirrup spacing of the structural columns were obtained, namely 142, 158, 135, 165, 148, 172, 139, 155, 161, 144, 167, 153, 138, 163, 149, 171, 141, 157, 164, 146, 168, 154, 137, 162, 151, 169, 143, 159, 166, 147, 152, and 156, all in millimeters. The sample mean was calculated to be 154.4 mm, and the sample standard deviation was 10.3 mm. A normal distribution was used for fitting, with a mean parameter μ of 154.4 and a standard deviation parameter σ of 10.3. The Anderson-Darling statistic after fitting was 0.42, which is less than the critical value of 0.75 corresponding to a significance level of 0.05, indicating that the normal distribution fit is acceptable. The distribution type (normal), mean μ = 154.4, standard deviation σ = 10.3, and Anderson-Darling statistic = 0.42 were stored in the stirrup spacing record of the regional prior structural feature database. The quantile function of this record is: for any given probability value p∈(0, 1), quantile Q(p) = μ + σ × Φ -1(p), where Φ -1 This is the inverse cumulative distribution function of the standard normal distribution. For example, Figure 3 is an example data table of regional structural construction quality parameters provided in an embodiment of the present invention, showing the original collected data of some buildings among 32 sample buildings selected in the target area.

[0049] The following technical effects were achieved through this step:

[0050] This step leverages the statistical commonalities in construction practices and material supply among self-built houses in the same region and period. By selecting more than 30 sample buildings for on-site and historical data collection, sample data on hidden structural parameters and directly measurable parameters are obtained. Probability distribution fitting is performed using distribution types that match the characteristics of each parameter value, constructing a regional prior structural feature database characterized by probability distributions. This transforms the problem of scarce individual data into a regional statistical inference problem.

[0051] S200: Based on the building structure data acquisition scheme for the target building, the construction quality parameters are divided into identifiable categories.

[0052] A set of fixed parameters and a set of uncalibrated parameters are used, and combined with the historical building construction data, a state quantile transition matrix is ​​constructed accordingly.

[0053] The regional prior structural feature database quantifies the statistical regularities of construction quality parameters for similar buildings within the target region in the form of probability distributions. When creating a digital twin model of a target building, it is necessary to obtain the specific structural parameter values ​​of the building as model input. However, the difficulty of obtaining different structural parameters varies significantly: parameters such as building plan dimensions, wall thickness, and floor height can be directly measured using conventional tools such as measuring tapes or laser rangefinders; parameters such as mortar compressive strength, masonry method, actual spacing of stirrups inside structural columns, and the setting of tie bars between walls and structural columns are hidden inside the masonry and concrete and cannot be directly obtained through non-destructive testing methods.

[0054] Step S200 provided in this embodiment of the invention includes:

[0055] A data acquisition scheme for obtaining the building structure data of the target building;

[0056] Analyze the building structure data acquisition scheme to determine the accessibility of detection methods corresponding to each structural construction quality parameter;

[0057] Based on the accessibility, the construction quality parameters are divided into a set of calibrable parameters that can be directly obtained through non-destructive testing and a set of non-calibrable parameters that cannot be directly obtained through non-destructive testing.

[0058] The specific implementation method is as follows:

[0059] First, a data collection plan for the building's structural features is obtained. This plan is prepared by a qualified building structure testing and assessment agency before conducting on-site surveys. The plan is based on national standards, the building's construction date and structural type, and common construction practices for self-built houses in local villages and towns. Once completed, the plan is delivered to the data collection team in the form of a formal survey technical document. The plan typically lists all planned structural quality parameters to be collected, the corresponding testing methods and equipment models for each parameter, and the expected accuracy level for each parameter in tabular form. During the on-site survey, the team collects parameters from the target building item by item according to the technical route specified in the plan.

[0060] Then, the building structure data acquisition scheme is analyzed to determine the accessibility of the detection methods corresponding to each structural construction quality parameter. Accessibility is determined based on the physical accessibility of the detection method to the target parameter measurement location and the destructive nature of the measurement process. Parameters that can be directly obtained through non-destructive or minimal-destructive testing methods such as measuring tapes, laser rangefinders, rebar scanners, and rebound hammers are deemed calibrable. Parameters that must be obtained through destructive methods such as partial chiseling or sampling tests are deemed uncalibrable.

[0061] Based on the accessibility assessment results, all construction quality parameters are divided into a calibrable parameter set and a non-calibrable parameter set. Parameters in the calibrable parameter set can be directly obtained through non-destructive testing, and typical parameter types include building plan dimensions, floor height, wall thickness, structural column cross-sectional dimensions, floor slab type and thickness, and number of floors. Parameters in the non-calibrable parameter set cannot be directly obtained through non-destructive testing, and typical parameter types include mortar compressive strength and masonry shear strength.

[0062] Strength, masonry method, mortar fullness, spacing of stirrups in structural columns, stirrup diameter, and spacing of tie bars.

[0063] Step S200 provided in this embodiment of the invention further includes:

[0064] Based on multiple probability distributions in the prior structural feature database of the region, the continuous value space of each structural construction quality parameter in the historical building structural data is discretized into a finite number of quantile states.

[0065] Based on the relationship between the calibrable and non-calibrable parameters according to the construction quality parameters, and using each data entry in the historical building construction data as a joint observation index, a corresponding state quantile transition matrix is ​​constructed for each non-calibrable parameter.

[0066] The elements of the state quantile transition matrix are: the conditional probability of the uncalibrable parameter being in each quantile state when the calibrable parameter is in any quantile state.

[0067] The specific implementation method is as follows:

[0068] Based on multiple probability distributions in the regional prior structural feature database, the continuous value space of each structural construction quality parameter in the historical building construction data is discretized into a finite number of quantile states. The number of quantile states is preset to 4. The intervals for the 4 quantile states are as follows: the 0th to 25th percentile is state 1, the 25th to 50th percentile is state 2, the 50th to 75th percentile is state 3, and the 75th to 100th percentile is state 4. Each parameter is associated with a quantile function, which is obtained from the regional prior structural feature database.

[0069] The discretization process is as follows: For each data record in the historical building construction data, the original continuous value of each construction quality parameter in that record is read. The corresponding quantile function is then called to convert the original continuous value into a corresponding quantile value p. The quantile value p represents the relative position of the parameter value in the prior distribution of the region, ranging from 0 to l. The quantile state is determined based on the interval in which p falls: state 1 is when 0 ≤ p < 0.25, state 2 is when 0.25 ≤ p < 0.5, state 3 is when 0.5 ≤ p < 0.75, and state 4 is when 0.75 ≤ p ≤ l. After all historical data records are discretized, each construction quality parameter in each data record is replaced with its corresponding quantile state number.

[0070] Then, based on the attribution relationship between calibrable and non-calibrable parameters according to the construction quality parameters, and using each data entry in the historical building construction data as a joint observation index, a corresponding state quantile transition matrix is ​​constructed for each non-calibrable parameter. For a certain non-calibrable parameter Y and a certain calibrable parameter X, the corresponding state quantile transition matrix M is a 4-row, 4-column matrix. The element M(i,j) in the i-th row and j-th column of the matrix represents the conditional probability that the non-calibrable parameter Y is in the j-th quantile state when the calibrable parameter X is in the i-th quantile state.

[0071] The conditional probability is calculated using data entries from historical building construction data that simultaneously contain complete observations of both calibrable parameter X and uncalibrable parameter Y as a joint observation sample. The calculation steps are as follows: traverse all data entries in the joint observation sample and count the total number of data entries where the calibrable parameter X is in the i-th quantile state, denoted as N_X(i). Among the N_X(i) data entries where the calibrable parameter X is in the i-th quantile state, further count the number of data entries where the uncalibrable parameter Y is simultaneously in the j-th quantile state, denoted as N_XY(i,j). The conditional probability M(i,j) = N_XY(i,j) ÷ N_X(i). (State quantiles)

[0072] Each row of the number transfer matrix satisfies the probability normalization constraint, that is, the sum of the elements in each column of that row is l.

[0073] Specifically, based on the attribution relationship between the calibrable and non-calibrable parameters according to the construction quality parameters, and using each data entry in the historical building construction data as a joint observation index, a corresponding state quantile transition matrix is ​​constructed for each non-calibrable parameter, including:

[0074] Principal component analysis is performed on the historical building construction data. For each uncalibrable parameter, multiple strongly correlated calibrable parameters are selected and determined, and the normalized principal component contribution of each strongly correlated calibrable parameter is calculated.

[0075] For each strongly correlated calibrable parameter, statistical analysis is performed based on the historical building construction data under quantile state conditions to obtain the conditional statistical distribution of the uncalibrable parameter under each quantile state for each strongly correlated calibrable parameter.

[0076] For each binary parameter combination consisting of an uncalibrable parameter and a strongly correlated calibrable parameter, a single-association parameter mapping is established to obtain the state quantile transition matrix of the single association corresponding to each binary parameter combination.

[0077] The state quantile transition matrix, the corresponding principal component contribution, and the corresponding binary parameter combination label are associated, stored, and output.

[0078] The specific implementation method is as follows:

[0079] Principal component analysis (PCA) is a data dimensionality reduction and feature extraction method. Its principle is to transform potentially correlated variables into linearly uncorrelated principal components through orthogonal transformation. The principal components are arranged in descending order of their variance contribution rate, with the first few principal components typically retaining most of the variance information of the original data. When performing PCA on historical building construction data, all calibrable parameters are used as the original variables to construct an original data matrix. Each row of the matrix corresponds to a historical data record, and each column corresponds to the standardized value of a calibrable parameter. The covariance matrix is ​​calculated, and eigenvalue decomposition is performed to obtain each eigenvalue and its corresponding eigenvector. The eigenvalues ​​are arranged in descending order, and the eigenvector corresponding to each eigenvalue represents the loading direction of a principal component. The variance contribution rate of the k-th principal component is equal to that eigenvalue divided by the sum of all eigenvalues.

[0080] For a given uncalibrable parameter Y, calculate its Pearson correlation coefficient with each principal component. The Pearson correlation coefficient is calculated by dividing the covariance between the two variables by the product of their respective standard deviations, and its value ranges from -1 to 1. The larger the absolute value, the stronger the linear correlation. Select the top 3 principal components with the largest absolute values ​​of their correlation coefficients with the uncalibrable parameter Y. From these 3 principal components, extract the calibrable parameter with the largest absolute value of its loading coefficient, and denote it as the strongly correlated calibrable parameter of Y, denoted as X1, X2, and X3, respectively.

[0081] The absolute values ​​of the load coefficients corresponding to the three strongly correlated calibrable parameters are normalized so that the sum of the normalized contributions is l. The normalization method is: w k =|loading k |÷(|loading1|+|loading2|+|loading3|), where k takes values ​​of 1, 2, and 3, loading k The loading coefficients of the k-th strongly correlated parameter on its principal component. k This is the normalized principal component contribution corresponding to the strongly correlated calibrable parameter.

[0082] For each strong association, the parameter X can be calibrated. k Statistical analysis under quantile conditions is performed based on historical building construction data. Following the aforementioned method for calculating conditional probability, X is constructed respectively. k The single-correlation state quantile transition matrix M between the uncalibrated parameter Y and the parameter Y k The quantile transition matrix M for each single-association state k It is a 4x4 grid with element M. k (i,j) represents the strongly correlated calibrable parameter X.k When in the i-th quantile state, the conditional probability that the unlabelable parameter Y is in the j-th quantile state.

[0083] The single-association state quantile transition matrix M corresponding to each strongly correlated calibrable parameter. k Normalized principal component contribution w k With the corresponding binary parameter combination (X) k The associated storage output is marked as Y). In subsequent steps, for the inference of a certain unlabelable parameter, the corresponding single-association state quantile transition matrix will be queried according to the quantile state of each of the multiple strongly associated labelable parameters. The conditional probability distributions of each matrix will be weighted and fused with the contribution of each normalized principal component as the weight to obtain the comprehensive conditional probability distribution.

[0084] The following technical effects were achieved through this step:

[0085] First, this step uses the accessibility of detection methods as an objective criterion to divide the construction quality parameters into two categories: calibrable and non-calibrable, thus clarifying the parameter acquisition path in digital twin modeling. Based on the probability distribution of the regional prior construction feature database, the parameters are discretized using quantiles, mapping the continuous value space to a unified finite quantile state space.

[0086] Second, using joint observation entries from historical building construction data as samples, a state quantile transition matrix was constructed from quantile states of calibrable parameters to quantile states of non-calibrable parameters. Principal component analysis was used to select the multiple calibrable parameters most strongly correlated with non-calibrable parameters. A single-association state quantile transition matrix was established for each strongly correlated calibrable parameter, and normalized contribution weights were calculated. This allows the inference of non-calibrable parameters to incorporate information from multiple calibrable parameters.

[0087] S300: Obtain the measured calibration parameter values ​​of each calibrable parameter of the target building, and map them to the corresponding quantile states based on the prior construction of the feature database of the region.

[0088] In the parameter acquisition process of digital twin modeling, calibrable parameters, which can be directly obtained through non-destructive testing, serve as the entry data for initiating the inference process of non-calibrable parameters. The measured values ​​of calibrable parameters are continuous numerical values, while the indices in the state quantile transition matrix are based on the discretized state space of quantile states. Therefore, it is necessary to map the measured values ​​of each calibrable parameter to a unified quantile state number through the probability distribution and quantile function in the regional prior feature database. This establishes a standardized mapping interface from the continuous measured values ​​of individual buildings to the regional statistical discrete states, providing a standardized input format for subsequent queries of the state quantile transition matrix using quantile states as indexes.

[0089] Step S300 provided in this embodiment of the invention includes:

[0090] On-site surveys were conducted on the target building, and the measured calibration parameter values ​​of each calibrable parameter were collected using non-destructive testing methods.

[0091] Match the probability distribution and its quantile function corresponding to each calibrable parameter in the prior construction feature database of the region;

[0092] Based on the probability distribution and the quantile function, each of the measured calibration parameter values ​​is mapped to the corresponding quantile state.

[0093] The specific implementation method is as follows:

[0094] A site survey was conducted on the target building, and non-destructive testing (NDT) methods were used to collect measured values ​​of various calibrable parameters. NDT refers to testing techniques that directly acquire parameter values ​​using physical measuring instruments without causing structural damage to the target building. For building plan dimensions, a laser rangefinder was used to measure the dimensions in each bay and depth direction. The laser rangefinder emits laser pulses and receives reflected signals, calculating the distance based on the laser's round-trip time. The measurement accuracy is down to the millimeter level, outputting the measured values ​​of bay and depth dimensions in millimeters. For floor height, a laser rangefinder was used to measure vertically between the bottom and top surfaces of each floor slab, outputting the measured values ​​of each floor height in millimeters. For wall thickness, a steel ruler or vernier caliper was used to directly measure at door and window openings, outputting the measured values ​​of wall thickness in millimeters. For structural column cross-sectional dimensions, a steel ruler was used to measure the width and depth of the exposed surface of the structural column, outputting the measured values ​​of the structural column cross-sectional width and depth in millimeters. For floor slab type and thickness, the measured floor slab thickness is obtained by observing the shape of the bottom surface of the floor slab and measuring the thickness of the exposed parts of the floor slab edges, and is output as the actual measured value in millimeters. For the number of floors, the value is directly counted by visual observation and output.

[0095] For example, a site survey was conducted on a target building. The building is a two-story brick-concrete structure. The dimensions of the bays were measured using a laser rangefinder to be 3600mm, 3300mm, and 3600mm, and the dimensions of the depths were 4200mm and 4500mm, respectively. The floor heights were measured to be 3300mm for the first floor and 3000mm for the second floor. The wall thickness was measured to be 240mm. The cross-sectional dimensions of the structural columns were measured to be 240mm in width and 240mm in depth. The floor slab thickness was measured to be 120mm. The building has two floors.

[0096] The probability distributions and quantile functions corresponding to each calibrable parameter are matched in the regional prior structural feature database. The database stores the distribution type, distribution parameter, and quantile function for each structural construction quality parameter, using the parameter name as the primary key. The matching method involves using the parameter name as the query key to perform an exact search in the database, returning the corresponding distribution type, distribution parameter, and quantile function. For calibrable parameters that do not have an independent record in the regional prior structural feature database, the probability distribution of the parameter in the database that is closest to its physical meaning is used as a substitute.

[0097] Based on probability distributions and quantile functions, each measured calibration parameter value is mapped to its corresponding quantile state. The quantile function is the inverse cumulative distribution function of the parameter's probability distribution. Its function is to return the position of a given continuous value within the probability distribution, denoted by the quantile value p, which ranges from 0 to 1. The quantile value p represents the relative position of the measured value within the regional prior distribution: p = 0.5 indicates that the measured value is exactly at the median level of the regional distribution; p < 0.5 indicates that the measured value is below the typical level of the region; and p > 0.5 indicates that the measured value is above the typical level of the region.

[0098] The quantile value p is converted according to the four preset quantile state intervals in step S200 to obtain the corresponding quantile state number. The division rule of the four state intervals is as follows: when 0 ≤ p < 0.25, the quantile state is 1; when 0.25 ≤ p < 0.5, the quantile state is 1.

[0099] When 0.5 ≤ p < 0.75, the quantile state is 2; when 0.5 ≤ p < 0.75, the quantile state is 3; when 0.75 ≤ p ≤ 1, the quantile state is 4. After mapping the measured values ​​of all calibrable parameters, the quantile state number corresponding to each calibrable parameter is obtained.

[0100] For example, for the measured wall thickness of the target building mentioned above, which is 240 mm, the regional prior structural feature database matches a wall thickness that follows a normal distribution with a mean μ = 240 mm and a standard deviation σ = 15 mm. The quantile function is the inverse cumulative distribution function of the standard normal distribution after location and scale transformation: Q(p) = μ + σ·Φ -1 (p), where Φ -1This is the inverse cumulative distribution function of the standard normal distribution. Substituting the measured value of 240mm into the equation: p = Φ((240-240) / 15) = Φ(0) = 0.50. This quantile p = 0.50 falls within the interval 0.50 ≤ p < 0.75, and is mapped to quantile state 3. For the measured floor height of 3300mm, it is matched to a floor height that follows a normal distribution with a mean μ = 3000mm and a standard deviation σ = 120mm. Calculating p = Φ((3300-3000) / 120) = Φ(2.50) ≈ 0.994, this is mapped to quantile state 4. After completing the mapping of all calibrable parameters, each calibrable parameter is converted into a quantile state number between 1 and 4, forming the quantile state vector of the calibrable parameters.

[0101] The following technical effects were achieved through this step:

[0102] By collecting measured values ​​of various calibrable parameters of the target building through on-site surveys, and utilizing the probability distribution and quantile function of each parameter in the regional prior construction feature database, continuous measured values ​​of different dimensions and value ranges are uniformly mapped to quantile state numbers from 1 to 4. This mapping expresses the locational characteristics of the measurable parameters of an individual building in a unified discrete state coding form, establishing a standardized interface from the continuous measured values ​​of an individual building to the regional statistical discrete state.

[0103] S400: Using the quantile states of the calibrable parameters as indexes, query the state quantile transition matrix to obtain the conditional probability distribution of each uncalibrable parameter in each quantile state, and define the predicted calibration parameter values ​​of each uncalibrable parameter based on the conditional probability distribution.

[0104] Step S400 provided in this embodiment of the invention includes:

[0105] Using the quantile states of each calibrable parameter as indexes, query the rows of each uncalibrable parameter in the corresponding state quantile transition matrix, and extract the conditional probability distribution of each uncalibrable parameter in each quantile state.

[0106] Based on the multiple conditional probability distributions, calculate and obtain the set of expected quantile values ​​for each uncalibrated parameter;

[0107] Using each uncalibrable parameter as an index and the normalized principal component contribution associated with the state quantile transition matrix as a weight, the expected quantile values ​​corresponding to multiple strongly correlated calibrable parameters are weighted and fused to obtain the predicted quantile of each uncalibrable parameter.

[0108] In the prior construction feature database of the region, the probability distribution and quantile function corresponding to each uncalibrated parameter are matched, and each predicted quantile is reverse-mapped and restored to obtain the predicted calibration parameter value of each uncalibrated parameter.

[0109] The specific implementation method is as follows:

[0110] First, using the quantile states of each calibrable parameter as indexes, query the corresponding states of each uncalibrable parameter.

[0111] From the rows of the quantile transition matrix, extract the conditional probability distribution of each unlabelable parameter at each quantile state. For a given unlabelable parameter Y, the above steps have selected three strongly correlated labelable parameters X1, X2, and X3, and constructed corresponding single-correlation state quantile transition matrices M1, M2, and M3, respectively. Each transition matrix M... k Given a 4x4 matrix, the i-th row represents the strongly correlated calibrable parameter X. k The conditional probability distribution of the unlabelable parameter Y in each quantile state when it is in the i-th quantile state. The three strongly correlated labelable parameters are each mapped to their corresponding quantile state numbers. Using the quantile state number i1 of X1 as the row index, all four elements of the i1-th row are extracted from M1. These four elements constitute the conditional probability distribution of the unlabelable parameter Y based on X1, denoted as P1(j) = M1(i1,j), where j = 1, 2, 3, 4. Similarly, using the quantile state number i2 of X2 as the row index, the conditional probability distribution P2(j) = M2(i2,j) of the i2-th row is extracted from M2. Using the quantile state number i3 of X3 as the row index, the conditional probability distribution P3(j) = M3(i3,j) of the i3-th row is extracted from M3.

[0112] Then, based on multiple conditional probability distributions, the set of expected quantile values ​​for each uncalibrated parameter is calculated. For the conditional probability distribution P1(j) based on X1, the expected quantile value is calculated as a weighted sum of the central quantile value of each quantile state and the probability of that state: E1=Σ(j=1,2,3,4)P1(j)×qj, where qj is the central quantile value of the j-th quantile state. The central quantile values ​​for the four quantile states are taken as the midpoint quantile values ​​of each state interval: the central quantile value of state 1 is q1 = 0.125, corresponding to the midpoint of the 0th to 25th percentile; the central quantile value of state 2 is q2 = 0.375, corresponding to the midpoint of the 25th to 50th percentile; the central quantile value of state 3 is q3 = 0.625, corresponding to the midpoint of the 50th to 75th percentile; and the central quantile value of state 4 is q4 = 0.875, corresponding to the midpoint of the 75th to 100th percentile. Similarly, the expected quantile value E2 based on condition X2 is calculated as E2 = Σ(j=1,2,3,4)P2(j)×qj, and the expected quantile value E3 based on condition X3 is calculated as E3 = Σ(j=1,2,3,4)P3(j)×qj. The units of E1, E2, and E3 are all quantiles, meaning that the range of all three expected values ​​is between 0 and 1. The three quantile expected values ​​are denoted as the set of quantile expected values ​​{E1, E2, E3} of the uncalibrated parameter Y.

[0113] Next, using each unlabelable parameter as an index and the normalized principal component contribution associated with the state quantile transition matrix as weight, the expected quantile values ​​corresponding to multiple strongly correlated labelable parameters are weighted and fused. The normalized principal component contributions w1, w2, and w3 are obtained from the principal component analysis process in step S200, satisfying w1 + w2 + w3 = 1. The comprehensive predicted quantile E of the weighted fusion is calculated as w1 × E1 + w2 × E2 + w3 × E3, and the result is a quantile value between 0 and 1. The role of the normalized principal component contribution is to measure the relative magnitude of the statistical explanatory power of each strongly correlated labelable parameter for the unlabelable parameter. The larger the contribution of the labelable parameter, the greater the weight of its corresponding expected quantile value in the weighted fusion.

[0114] After weighted fusion, the predicted quantile E for each unlabelable parameter is obtained. The predicted quantile E is a weighted average quantile value that combines the independent inference results of multiple strongly correlated labelable parameters, indicating the position of the unlabelable parameter in the prior distribution of the region.

[0115] Finally, the probability distribution and quantile function corresponding to each unlabelable parameter are matched in the regional prior feature database, and inverse mapping is performed on each predicted quantile. The inverse mapping method is as follows: substitute the predicted quantile E into the quantile function Q(p) corresponding to the unlabelable parameter, calculate Q(E), and the result is the predicted quantile of the unlabelable parameter.

[0116] l3

[0117] The parameter values ​​are fixed, and the units are consistent with the original parameters. The quantile function Q(p) is calculated and stored during the construction of the regional prior feature database, and is the inverse cumulative distribution function of the corresponding probability distribution.

[0118] For example, taking the non-calibrable parameter of stirrup spacing in a structural column as an example. Step S200 has selected three strongly correlated calibrable parameters: wall thickness X1, structural column cross-sectional width X2, and number of stories X3. Step S300 has mapped these three calibrable parameters to quantile states respectively: wall thickness as state 3, structural column cross-sectional width as state 2, and number of stories as state 4. Step S200 has constructed single-correlation state quantile transition matrices M1, M2, and M3 between each strongly correlated calibrable parameter and stirrup spacing, and calculated the normalized principal component contributions: w1=0.40, w2=0.35, and w3=0.25. Using the quantile state 3 of the wall thickness as the row index, the conditional probability distribution of the 3rd row is extracted from M1: P1(l)=0.05, P1(2)=0.12, P1(3)=0.38, P1(4)=0.45. The expected value of the quantile based on the wall thickness is E1=0.05×0.125+0.12×0.375+0.38×0.625+0.45×0.875=0.6825.

[0119] Using the quantile state 2 of the column cross-section width as the row index, extract the conditional probability distribution of the second row in M2:

[0120] P2(l)=0.08, P2(2)=0.22, P2(3)=0.42, P2(4)=0.28. The expected value of the quantile based on the width of the structural column section is E2=0.08×0.125+0.22×0.375+0.42×0.625+0.28×0.875=0.60.

[0121] Using the quantile state 4 of the layer as the row index, the conditional probability distribution of the 4th row is extracted in M3: P3(1)=0.02, P3(2)=0.05, P3(3)=0.28, P3(4)=0.65. The expected value of the quantile based on the layer is E3=0.02×0.125+0.05×0.375+0.28×0.625+0.65×0.875=0.765.

[0122] The weighted fusion comprehensive prediction quantile E = 0.40 × 0.6825 + 0.35 × 0.60 + 0.25 × 0.765 = 0.67425, approximately 0.674.

[0123] The regional prior structural feature database constructed in the above steps records that the stirrup spacing follows a normal distribution, with a mean μ = l54.4 mm and a standard deviation σ = l0.3 mm. Substituting the predicted quantile E = 0.674 into the quantile function, and referring to the table, we get Φ_1(0.674)≈0.45, Q(0.674) = l54.4 + l0.3 × 0.45 = l54.4 + 4.635≈l59.0 mm. Therefore, the predicted calibration parameter value for the stirrup spacing of the structural column is l59 mm.

[0124] The following technical effects were achieved through this step:

[0125] This step uses the quantile states of calibrable parameters as indices to extract conditional probability distributions from the quantile transition matrices of single-association states corresponding to multiple strongly correlated calibrable parameters. Weighted fusion is then performed using normalized principal component contributions, achieving comprehensive inference of uncalibrable parameters based on multi-source measurable information. The predicted value of each uncalibrable parameter is a weighted synthesis of the independent inference results of multiple measurable parameters according to their respective statistical interpretability. Measurement errors of individual calibrable parameters or biases of individual samples do not dominate the final inference result.

[0126] S500: Combining the measured calibration parameter values ​​and the predicted calibration parameter values, a digital twin building for the target building is constructed. Under the set seismic loading conditions, structural response simulation is performed, and seismic toughness monitoring results are obtained based on the corresponding evaluation of the structural response simulation results.

[0127] The measured calibration parameter values ​​provide information on the overall geometric dimensions and mass distribution of the target building, and the predicted calibration parameter values ​​provide...

[0128] It provides concealed structural parameters such as mortar strength, masonry method, stirrup spacing, and tie bar placement. Digital twin buildings need to integrate these two types of parameters into a structural analysis model within a unified mechanical framework to simulate the damage and deformation response of each component under different earthquake intensities, and evaluate the seismic toughness of the target building from four dimensions: structural damage resistance, force transmission path redundancy, functional recovery capability, and availability of repair resources.

[0129] Step S500 provided in this embodiment of the invention includes:

[0130] Based on the measured calibration parameter values, a geometric model and a mass distribution model of the target building are constructed.

[0131] Based on the predicted calibration parameter values, a mechanical model of masonry wall panels, a structural column-ring beam constraint effect model, and a wall-structural column tie bar connection model are constructed respectively.

[0132] The geometric model, the mass distribution model, the masonry wall slab mechanical model, the structural column-ring beam constraint effect model, and the wall-structural column tie bar connection model are assembled to construct the digital twin of the target building.

[0133] The seismic loading conditions are set, and the structural response of the digital twin building is simulated to obtain the seismic response results of each component.

[0134] The specific implementation method is as follows:

[0135] First, based on measured calibration parameters, a geometric model and a mass distribution model of the target building are constructed. The geometric model describes the building's spatial configuration and component arrangement. It takes measured calibration parameters such as building plan dimensions, floor height, wall thickness, and structural column cross-sectional dimensions as input, and establishes a three-dimensional geometric wireframe of the structure in the preprocessing module of the finite element analysis software. Each floor's walls are arranged into several wall segments according to the measured bay and depth dimensions, with the wall segment thickness taken from the measured wall thickness. Floor elevations are set between floors according to the measured floor height. Structural columns are arranged at the intersections of longitudinal and transverse walls and at the corners of external walls, according to the measured cross-sectional dimensions. Ring beams are continuously arranged along the top of each floor's walls. The mass distribution model describes the representative values ​​of gravity loads on each floor. The total mass of each floor is calculated based on the measured floor slab thickness and the standard value of the dead load on the brick-concrete structure floor, concentrating the mass distribution at each floor elevation to form the structure's mass matrix. The standard value of the floor dead load is taken from the literature recommendations, and the live load is reduced according to the standard GB 50009.

[0136] Based on the predicted calibration parameter values, mechanical models of masonry wall panels, constraint effect models of structural column ring beams, and connection models of tie bars of structural columns in walls were constructed respectively.

[0137] The mechanical model of the masonry wall uses the mortar compressive strength and masonry method from the predicted calibration parameters as input mechanical parameters. The mortar compressive strength is directly used as the calculation parameter for the masonry compressive strength. The masonry compressive strength is calculated from the mortar compressive strength and brick compressive strength using an empirical formula: fm = k1 × fb^α × f_mortar^β, where fm is the average masonry compressive strength in MPa, fb is the average brick compressive strength in MPa, f_mortar is the predicted mortar compressive strength in MPa, and k1, α, and β are empirical coefficients. The elastic modulus of the masonry is taken as an empirical multiple of the masonry compressive strength. The stress-strain relationship of the masonry adopts a two-segment constitutive model with a parabolic ascending segment and a linear descending segment. The masonry wall is discretized using layered shell elements, which consist of multiple material layers, each assigned a masonry material and corresponding thickness.

[0138] The structural column-ring beam constraint effect model uses the stirrup spacing and diameter from the predicted calibration parameters as constraint parameter inputs. The constraint effect of the structural column on the masonry wall is quantified by the constraint coefficient η, η = 1 + ρv × fyv × bc / fm, where ρv is the volumetric stirrup ratio, ρv equals the stirrup cross-sectional area multiplied by the stirrup perimeter divided by the stirrup spacing multiplied by the core area, fyv is the stirrup yield strength, bc is the structural column cross-section width, and fm is the masonry compressive strength. The predicted values ​​of stirrup spacing and diameter are used to calculate the volumetric stirrup ratio. The masonry compressive strength after constraint is increased to f'm = η × fm, and the ultimate compressive strain after constraint is correspondingly increased. The structural column and ring beam are simulated using fiber beam elements. The cross-section of the fiber beam element is divided into several fibers, and each fiber is assigned a uniaxial constitutive relation to the steel reinforcement or concrete.

[0139] The connection model for the tie bars in the wall and structural columns uses the tie bar spacing and diameter from the predicted calibration parameters as input connection parameters. The equivalent connection stiffness of the tie bars in the collaborative action between the wall and the structural column is simulated using a series spring model. The tensile bearing capacity of a single tie bar is Tr = As × fy, where As is the cross-sectional area of ​​a single tie bar, and fy is the yield strength of the steel reinforcement. Spring elements are arranged on the interface between the wall and the structural column, with the spring axis perpendicular to the wall plane. The number of tie bars represented by each spring element is determined based on the predicted tie bar spacing and the wall segment length. The yield bearing capacity of the spring is equal to the number of tie bars represented by the spring multiplied by the tensile bearing capacity of a single tie bar, and the initial stiffness of the spring is equal to the elastic modulus of the steel reinforcement multiplied by the total area divided by the anchorage length.

[0140] Then, the geometric model, mass distribution model, masonry wall slab mechanical model, structural column-ring beam constraint effect model, and wall-structural column tie bar connection model are assembled in finite element analysis software. The assembly method involves assigning the masonry wall slab mechanical model to the layered shell elements corresponding to each wall segment in the geometric model, assigning the structural column-ring beam constraint effect model to the fiber beam elements corresponding to each structural column and ring beam, and assigning the wall-structural column tie bar connection model to the spring elements at the interface between the wall and the structural column. After assembly, a complete digital twin of the target building is obtained.

[0141] Furthermore, seismic loading conditions were set up to simulate the structural response of the digital twin building. The seismic loading conditions were determined based on the seismic fortification intensity and site category of the target building's location. Two seismic loading levels, namely fortification earthquake and rare earthquake, were selected for simulation. The ground motion input used artificial seismic wave acceleration time history generated from the site-related response spectrum, with a total time history duration of 30 seconds and a time step of 0.02 seconds. The structural response simulation employed a nonlinear dynamic time history analysis method, solving the dynamic equilibrium equations of the structure at each time step and outputting the nodal displacement, nodal velocity, and nodal acceleration response of each element at each time step.

[0142] After the simulation is completed, the seismic response results of each component are extracted. For the layered shell elements of the masonry wall panels, the maximum principal tensile strain and maximum principal compressive strain at each integration point are extracted; for the fiber beam elements of the structural columns and ring beams, the maximum tensile strain and maximum compressive strain of each fiber, as well as the sequence and time of each fiber entering the yield state in the cross-section, are extracted. The seismic response results of each component are then output as the structural response simulation results.

[0143] Step S500 provided in this embodiment of the invention further includes:

[0144] By using big data methods, we can dynamically acquire information on restoration resources in the target area;

[0145] Based on the results of the structural response simulation and the repair resource information, the structural damage resistance index, force transmission path redundancy index, functional recovery capability index, and repair resource availability index are calculated respectively.

[0146] The seismic toughness index is obtained by weighting the structural damage resistance index, the force transmission path redundancy index, the functional recovery capability index, and the repair resource availability index according to preset weights, and the output is the seismic toughness monitoring result.

[0147] The specific implementation method is as follows:

[0148] Big data methods are used to dynamically acquire restoration resource information for the target region. This information includes three categories: building material supply data, skilled worker data, and economic data. Building material supply data comes from real-time prices and inventory information in the local building materials market; skilled worker data comes from the number of bricklayers and steelworkers registered with the local construction authorities; and economic data comes from resident income levels and government post-disaster recovery budgets published by the local statistics department. Restoration resource information is updated regularly via an internet data interface. Building material market price indices and inventory data are typically released monthly, with an update cycle that can be preset to one month. Upon the occurrence of an earthquake, an immediate update is triggered to obtain the latest post-earthquake resource supply status.

[0149] Based on the results of structural response simulation and combined with repair resource information, the structural damage resistance index, force transmission path redundancy index, functional recovery capability index, and repair resource availability index are calculated respectively.

[0150] The structural damage resistance index reflects the degree to which a building resists damage under seismic loading. This index is calculated based on the weighted damage index of each component, which is calculated using the modified Park-Ang damage model. The modified Park-Ang damage model is calculated as D = δm / δu + β × dE / (Fy × δu), where δm is the maximum deformation of the component under seismic loading, obtained by extracting the maximum displacement of each node of the component from the structural response simulation results; δu is the ultimate deformation of the component under monotonic loading, determined by the statistical value of the component's quasi-static test data or the empirical value recommended by the code. For masonry walls, δu is taken as 1 / 150 of the wall height. β is the energy dissipation factor, reflecting the degree of degradation of the energy dissipation capacity of a component under seismic cyclic loading. For masonry walls, β is taken as 0.15. This value is based on the statistical results of quasi-static test data of masonry walls at home and abroad, and is an empirical value widely used in the literature in the modified Park-Ang damage model of masonry structures. dE is the cumulative hysteretic energy dissipation, obtained by the area integral of the force and displacement hysteresis curves of the component in the structural response simulation results. Fy is the yield bearing capacity of the component. For example, the weight coefficients of each component are set according to the component type: the weight coefficient of load-bearing walls is 1.0, the weight coefficient of structural columns is 0.85, the weight coefficient of ring beams is 0.6, and the weight coefficient of non-load-bearing partition walls is 0.3. The structural damage resistance index R1 = 1 / (1+D_avg), where D_avg is the arithmetic mean of the weighted damage indices of all components. The structural damage resistance index R1 ranges from 0 to 1. The closer R1 is to 1, the better the structural damage resistance.

[0151] The load transfer path redundancy index reflects the extent to which a building maintains its load-bearing capacity through alternative load transfer paths after the failure of some structural components. This index is calculated based on the proportion of effective lateral force-resisting components. The load-bearing walls are extracted from the structural response simulation results.

[0152] The inter-story drift angle is used to determine the effective lateral force resisting members. Load-bearing walls whose inter-story drift angle does not exceed the limit specified in GB 50011 for elasto-plastic inter-story drift angle of masonry structures are considered effective lateral force resisting members. The force transmission path redundancy index R2 = N_eff / N_total, where N_eff is the number of effective lateral force resisting members and N_total is the total number of load-bearing walls. The value of R2 ranges from 0 to 1; the closer R2 is to 1, the better the force transmission path redundancy.

[0153] The functional recovery capability index reflects the speed at which a building can regain its usability after an earthquake. This index is calculated based on the damage level assessment of its components. The damage index of each component is converted into a damage level according to a pre-defined damage level classification rule. Damage levels are divided into five levels: DS-1 is basically intact, corresponding to a damage index less than 0.1; DS-2 is slightly damaged, corresponding to a damage index between 0.1 and 0.3; DS-3 is moderately damaged, corresponding to a damage index between 0.3 and 0.6; DS-4 is severely damaged, corresponding to a damage index between 0.6 and 0.9; and DS-5 is collapsed or irreparable, corresponding to a damage index greater than 0.9. The highest damage level among all components is taken as the overall damage level of the building. The reference recovery times for each injury level are as follows: DS-1: 0 days; DS-2: 7 to 14 days; DS-3: 30 to 90 days; DS-4: 90 to 365 days; DS-5: more than 365 days. The median of the reference recovery time for each injury level is taken as the estimated recovery time T_rec. The functional recovery capability index R3 = exp(-T_rec / T_ref), where T_ref is the reference time scale, taken as 365 days. The value of R3 ranges from 0 to 1; the closer R3 is to 1, the faster the functional recovery.

[0154] The calculation of the repair resource availability index is based on the supply-demand ratio of building materials, skilled workers, and funds required for repair. Damage levels and repair workloads for each component are extracted from structural response simulation results, summarizing the overall building repair requirements: building material demand (D_material), skilled worker demand (D_labor), and funding demand (D_fund). The current supply capacity of building materials (S_material), skilled workers (S_labor), and funding (S_fund) are obtained from the target region's repair resource information. The repair resource availability index R4 = (S_material / D_material) × (S_labor / D_labor) × (S_fund / D_fund). In this formula, S_material / D_material is the building material supply-demand ratio, S_labor / D_labor is the skilled worker supply-demand ratio, and S_fund / D_fund is the funding supply-demand ratio. R4 ranges from 0 to 1; the closer R4 is to 1, the better the repair resource availability.

[0155] According to preset weights, the structural damage resistance index R1, the force transmission path redundancy index R2, the functional recovery capability index R3, and the repair resource availability index R4 are weighted and averaged to obtain the seismic toughness index R. The weighted average is calculated using a geometric weighted average: R = (R1^w1) × (R2^w2) × (R3^w3) × (R4^w4), where the sum of the four weight coefficients is 1. The seismic toughness index R is then output as the seismic toughness monitoring result. For example, the preset baseline values ​​for the four weighting coefficients are as follows: the weight of the structural damage resistance index is preset to 0.30. Self-built brick-concrete structures in villages and towns generally lack formal seismic design, resulting in significant dispersion in structural damage resistance; therefore, they should have a higher weight in the toughness evaluation. The weight of the force transmission path redundancy index is preset to 0.20. Two- to three-story brick-concrete structures in villages and towns are mainly supported by walls, resulting in relatively simple force transmission paths and small differences in force transmission path redundancy between buildings. Therefore, their distinguishing power in the toughness evaluation is limited, and a lower weight is assigned. The weight of the functional recovery capacity index is preset to 0.25. This index is closely related to the selection of repair schemes and the organization of repair work, and plays a crucial role in toughness evaluation.

[0156] This indicator has significant practical implications in the evaluation and is assigned a moderately high weight. The weight of the repair resource availability indicator is preset at 0.25. Post-earthquake recovery in rural areas often faces resource shortages, and even if the structural damage level is only moderate, the actual recovery time may be significantly prolonged if repair resources cannot be delivered in a timely manner. This indicator expands resilience assessment from a purely engineering dimension to a dimension that couples engineering and socio-economic factors, and has a substantial corrective effect on the resilience assessment results. Therefore, it is assigned the same weight as functional recovery capacity.

[0157] For target buildings located in remote mountainous areas or areas with inconvenient transportation, the weight of the repair resource availability index can be increased from the preset 0.25 to 0.40, the weight of the structural damage resistance index can be decreased to 0.25, the weight of the force transmission path redundancy index can be decreased to 0.15, and the weight of the functional recovery capability index can be decreased to 0.20, with the total remaining at 1, to reflect the decisive impact of resource availability on overall resilience in this scenario.

[0158] The following technical effects were achieved through this step:

[0159] This step constructs geometric and mass models using measured calibration parameters, and three mechanical sub-models—masonry wall panels, structural column ring beam constraints, and wall structural column tie bar connections—using predicted calibration parameters. Both calibrable and non-calibrable parameters are integrated in the digital twin building, and the seismic response of each component is obtained through nonlinear dynamic time history analysis. In the toughness assessment, based on the structural response simulation results and dynamically acquired repair resource information, a quantitative evaluation is conducted from four dimensions: structural damage resistance, force transmission path redundancy, functional recovery capability, and repair resource availability. A geometrically weighted average is used to comprehensively assess the seismic toughness level. The introduction of the repair resource availability dimension incorporates the actual resource constraints of post-earthquake recovery in rural areas into the toughness assessment, enabling the identification of risk scenarios that are technically repairable but lack the resources for timely restoration.

[0160] Example 2, as Figure 4 As shown, based on the same inventive concept as the digital twin-based dynamic monitoring method for seismic toughness of rural buildings provided in Embodiment 1, this embodiment of the invention also provides a digital twin-based dynamic monitoring system for seismic toughness of rural buildings, the system comprising:

[0161] The regional prior database construction module 11 is used to obtain historical building construction data of the target region corresponding to the target building, and extract regional construction quality parameters accordingly to construct a regional prior construction feature database based on probability distribution.

[0162] The parameter partitioning and migration matrix construction module 12 is used to divide the construction quality parameters into a calibrable parameter set and an uncalibrable parameter set according to the building construction data acquisition scheme of the target building, and construct the state quantile migration matrix in combination with the historical building construction data.

[0163] Quantile state mapping module 13 is used to obtain the measured calibration parameter values ​​of each calibrable parameter of the target building, and to map the feature database constructed based on the prior area to the corresponding quantile state.

[0164] The uncalibrable parameter inference module 14 is used to query the state quantile transition matrix with the quantile state of the calibrable parameter as the index, obtain the conditional probability distribution of each uncalibrable parameter in each quantile state, and define the predicted calibration parameter value of each uncalibrable parameter based on the conditional probability distribution.

[0165] The digital twin simulation and resilience evaluation module 15 is used to combine the measured calibration parameter values ​​with the predicted calibration parameters.

[0166] The system constructs a digital twin of the target building, simulates its structural response under a set seismic load condition, and obtains seismic toughness monitoring results based on the simulation results.

[0167] In one embodiment, the regional prior database construction module 11 is further used to select multiple buildings of the same structural type as the target building as sample buildings within the target region to which the target building belongs;

[0168] On-site and historical data of the structural construction quality characteristics of each sample building were collected to obtain hidden structural parameter samples and corresponding directly measurable parameter samples of each building, which were used as the structural construction quality parameters of the area.

[0169] For each of the hidden structural parameter samples and the corresponding directly measurable parameter samples, a probability distribution fitting is performed to obtain a regional prior structural feature database characterized by the probability distribution.

[0170] In one embodiment, the parameter partitioning and migration matrix construction module 12 is also used to obtain the building structure data acquisition scheme of the target building;

[0171] Analyze the building structure data acquisition scheme to determine the accessibility of detection methods corresponding to each structural construction quality parameter;

[0172] Based on the accessibility, the construction quality parameters are divided into a set of calibrable parameters that can be directly obtained through non-destructive testing and a set of non-calibrable parameters that cannot be directly obtained through non-destructive testing.

[0173] Based on multiple probability distributions in the prior structural feature database of the region, the continuous value space of each structural construction quality parameter in the historical building structural data is discretized into a finite number of quantile states.

[0174] Based on the relationship between the calibrable and non-calibrable parameters according to the construction quality parameters, and using each data entry in the historical building construction data as a joint observation index, a corresponding state quantile transition matrix is ​​constructed for each non-calibrable parameter.

[0175] The elements of the state quantile transition matrix are: the conditional probability of the uncalibrable parameter being in each quantile state when the calibrable parameter is in any quantile state.

[0176] Principal component analysis is performed on the historical building construction data. For each uncalibrable parameter, multiple strongly correlated calibrable parameters are selected and determined, and the normalized principal component contribution of each strongly correlated calibrable parameter is calculated.

[0177] For each strongly correlated calibrable parameter, statistical analysis is performed based on the historical building construction data under quantile state conditions to obtain the conditional statistical distribution of the uncalibrable parameter under each quantile state for each strongly correlated calibrable parameter.

[0178] For each binary parameter combination consisting of an uncalibrable parameter and a strongly correlated calibrable parameter, a single-association parameter mapping is established to obtain the state quantile transition matrix of the single association corresponding to each binary parameter combination.

[0179] The state quantile transition matrix, the corresponding principal component contribution, and the corresponding binary parameter combination label are associated, stored, and output.

[0180] In one embodiment, the quantile state mapping module 13 is also used to conduct on-site surveys of the target building and collect the measured calibration parameter values ​​of each calibrable parameter by means of non-destructive testing.

[0181] Match the probability distribution and its quantile function corresponding to each calibrable parameter in the prior construction feature database of the region;

[0182] Based on the probability distribution and the quantile function, each of the measured calibration parameter values ​​is mapped to the corresponding quantile state.

[0183] In one embodiment, the uncalibrable parameter inference module 14 is further configured to use the quantile states of each calibrable parameter as indexes to query the rows of each uncalibrable parameter in the corresponding state quantile transition matrix and extract the conditional probability distribution of each uncalibrable parameter in each quantile state.

[0184] Based on the multiple conditional probability distributions, calculate and obtain the set of expected quantile values ​​for each uncalibrated parameter;

[0185] Using each uncalibrable parameter as an index and the normalized principal component contribution associated with the state quantile transition matrix as a weight, the expected quantile values ​​corresponding to multiple strongly correlated calibrable parameters are weighted and fused to obtain the predicted quantile of each uncalibrable parameter.

[0186] In the prior construction feature database of the region, the probability distribution and quantile function corresponding to each uncalibrated parameter are matched, and each predicted quantile is reverse-mapped and restored to obtain the predicted calibration parameter value of each uncalibrated parameter.

[0187] In one embodiment, the digital twin simulation and resilience evaluation module 15 is also used to construct a geometric model and a mass distribution model of the target building based on the measured calibration parameter values;

[0188] Based on the predicted calibration parameter values, a mechanical model of masonry wall panels, a structural column-ring beam constraint effect model, and a wall-structural column tie bar connection model are constructed respectively.

[0189] The geometric model, the mass distribution model, the masonry wall slab mechanical model, the structural column-ring beam constraint effect model, and the wall-structural column tie bar connection model are assembled to construct the digital twin of the target building.

[0190] The seismic loading conditions are set, and the structural response of the digital twin building is simulated to obtain the seismic response results of each component.

[0191] By using big data methods, we can dynamically acquire information on restoration resources in the target area;

[0192] Based on the results of the structural response simulation and the repair resource information, the structural damage resistance index, force transmission path redundancy index, functional recovery capability index, and repair resource availability index are calculated respectively.

[0193] The seismic toughness index is obtained by weighting the structural damage resistance index, the force transmission path redundancy index, the functional recovery capability index, and the repair resource availability index according to preset weights, and the output is the seismic toughness monitoring result.

Claims

1. A method for dynamic monitoring of the seismic toughness of rural buildings based on digital twins, characterized in that, include: Obtain historical building construction data for the target region corresponding to the target building, extract regional construction quality parameters accordingly, and construct a regional prior construction feature database based on probability distribution; Based on the building construction data acquisition scheme of the target building, the construction quality parameters are divided into a set of calibrable parameters and a set of non-calibrable parameters, and a state quantile migration matrix is ​​constructed in conjunction with the historical building construction data. Obtain the measured calibration parameter values ​​of each calibrable parameter of the target building, and map them to the corresponding quantile states based on the prior features of the region. Using the quantile states of the calibrable parameters as indexes, query the state quantile transition matrix to obtain the conditional probability distribution of each uncalibrable parameter at each quantile state, and define the predicted calibration parameter values ​​of each uncalibrable parameter based on the conditional probability distribution. By combining the measured calibration parameter values ​​and the predicted calibration parameter values, a digital twin building for the target building is constructed. The structural response is simulated under the set seismic loading conditions, and the seismic toughness monitoring results are obtained based on the corresponding evaluation of the structural response simulation results.

2. The method for dynamic monitoring of seismic toughness of rural buildings based on digital twins as described in claim 1, characterized in that, Obtain historical building construction data for the target region corresponding to the target building, extract regional construction quality parameters accordingly, and construct a regional prior construction feature database based on probability distribution, including: Within the target area where the target building is located, select multiple buildings of the same structural type as the target building as sample buildings; On-site and historical data of the structural construction quality characteristics of each sample building were collected to obtain hidden structural parameter samples and corresponding directly measurable parameter samples of each building, which were used as the structural construction quality parameters of the area. For each of the hidden structural parameter samples and the corresponding directly measurable parameter samples, a probability distribution fitting is performed to obtain a regional prior structural feature database characterized by the probability distribution.

3. The method for dynamic monitoring of seismic toughness of rural buildings based on digital twins as described in claim 1, characterized in that, Based on the building structure data acquisition scheme for the target building, the construction quality parameters are divided into a set of calibrable parameters and a set of non-calibrable parameters, including: A data acquisition scheme for obtaining the building structure data of the target building; Analyze the building structure data acquisition scheme to determine the accessibility of detection methods corresponding to each structural construction quality parameter; Based on the accessibility, the construction quality parameters are divided into a set of calibrable parameters that can be directly obtained through non-destructive testing and a set of non-calibrable parameters that cannot be directly obtained through non-destructive testing.

4. The method for dynamic monitoring of seismic toughness of rural buildings based on digital twins as described in claim 3, characterized in that, Based on the aforementioned historical building construction data, a corresponding state quantile transition matrix is ​​constructed, including: Based on multiple probability distributions in the prior structural feature database of the region, the continuous value space of each structural construction quality parameter in the historical building structural data is discretized into a finite number of quantile states. Based on the relationship between the calibrable and non-calibrable parameters according to the construction quality parameters, and using each data entry in the historical building construction data as a joint observation index, a corresponding state quantile transition matrix is ​​constructed for each non-calibrable parameter. The elements of the state quantile transition matrix are: the conditional probability of the uncalibrable parameter being in each quantile state when the calibrable parameter is in any quantile state.

5. The method for dynamic monitoring of seismic toughness of rural buildings based on digital twins as described in claim 4, characterized in that, Based on the attribution relationship between the calibrable and non-calibrable parameters according to the construction quality parameters, and using each data entry in the historical building construction data as a joint observation index, a corresponding state quantile transition matrix is ​​constructed for each non-calibrable parameter, including: Principal component analysis is performed on the historical building construction data. For each uncalibrable parameter, multiple strongly correlated calibrable parameters are selected and determined, and the normalized principal component contribution of each strongly correlated calibrable parameter is calculated. For each strongly correlated calibrable parameter, statistical analysis is performed based on the historical building construction data under quantile state conditions to obtain the conditional statistical distribution of the uncalibrable parameter under each quantile state for each strongly correlated calibrable parameter. For each binary parameter combination consisting of an uncalibrable parameter and a strongly correlated calibrable parameter, a single-association parameter mapping is established to obtain the state quantile transition matrix of the single association corresponding to each binary parameter combination. The state quantile transition matrix, the corresponding principal component contribution, and the corresponding binary parameter combination label are associated, stored, and output.

6. The method for dynamic monitoring of seismic toughness of rural buildings based on digital twins as described in claim 1, characterized in that, Obtain the measured calibration parameter values ​​of each calibrable parameter of the target building, and construct a feature database based on the prior knowledge of the region to map it to the corresponding quantile state, including: On-site surveys were conducted on the target building, and the measured calibration parameter values ​​of each calibrable parameter were collected using non-destructive testing methods. Match the probability distribution and its quantile function corresponding to each calibrable parameter in the prior construction feature database of the region; Based on the probability distribution and the quantile function, each of the measured calibration parameter values ​​is mapped to the corresponding quantile state.

7. The method for dynamic monitoring of seismic toughness of rural buildings based on digital twins as described in claim 1, characterized in that, Using the quantile states of the calibrable parameters as indexes, the state quantile transition matrix is ​​queried to obtain the conditional probability distribution of each uncalibrable parameter at each quantile state. Based on the conditional probability distribution, the predicted calibration parameter values ​​for each uncalibrable parameter are defined, including: Using the quantile states of each calibrable parameter as indexes, query the rows of each uncalibrable parameter in the corresponding state quantile transition matrix, and extract the conditional probability distribution of each uncalibrable parameter in each quantile state. Based on the multiple conditional probability distributions, calculate and obtain the set of expected quantile values ​​for each uncalibrated parameter; Using each uncalibrable parameter as an index and the normalized principal component contribution associated with the state quantile transition matrix as a weight, the expected quantile values ​​corresponding to multiple strongly correlated calibrable parameters are weighted and fused to obtain the predicted quantile of each uncalibrable parameter. In the prior construction feature database of the region, the probability distribution and quantile function corresponding to each uncalibrated parameter are matched, and each predicted quantile is reverse-mapped and restored to obtain the predicted calibration parameter value of each uncalibrated parameter.

8. The method for dynamic monitoring of seismic toughness of rural buildings based on digital twins as described in claim 1, characterized in that, By combining the measured calibration parameter values ​​and the predicted calibration parameter values, a digital twin building for the target building is constructed, and structural response simulation is performed under a set seismic loading condition, including: Based on the measured calibration parameter values, a geometric model and a mass distribution model of the target building are constructed. Based on the predicted calibration parameter values, a mechanical model of masonry wall panels, a structural column-ring beam constraint effect model, and a wall-structural column tie bar connection model are constructed respectively. The geometric model, the mass distribution model, the masonry wall slab mechanical model, the structural column-ring beam constraint effect model, and the wall-structural column tie bar connection model are assembled to construct the digital twin of the target building. The seismic loading conditions are set, and the structural response of the digital twin building is simulated to obtain the seismic response results of each component.

9. The method for dynamic monitoring of seismic toughness of rural buildings based on digital twins as described in claim 1, characterized in that, Seismic toughness monitoring results are obtained based on the evaluation of structural response simulation results, including: By using big data methods, we can dynamically acquire information on restoration resources in the target area; Based on the results of the structural response simulation and the repair resource information, the structural damage resistance index, force transmission path redundancy index, functional recovery capability index, and repair resource availability index are calculated respectively. The seismic toughness index is obtained by weighting the structural damage resistance index, the force transmission path redundancy index, the functional recovery capability index, and the repair resource availability index according to preset weights, and the output is the seismic toughness monitoring result.

10. A dynamic monitoring system for the seismic toughness of rural buildings based on digital twins, characterized in that, Used to implement the claims The method for dynamic monitoring of seismic toughness of village and town buildings based on digital twins as described in any one of 1 to 9, wherein the system includes: a regional prior database construction module, used to acquire historical building structure data of the target region corresponding to the target building, and extract regional structural construction quality parameters accordingly, and construct a regional prior structural feature database based on probability distribution; The parameter partitioning and migration matrix construction module is used to divide the construction quality parameters of the target building into a set of calibrable parameters and a set of non-calibrable parameters according to the building construction data acquisition scheme, and to construct a state quantile migration matrix in combination with the historical building construction data. The quantile state mapping module is used to obtain the measured calibration parameter values ​​of each calibrable parameter of the target building, and to map the corresponding quantile state to the feature database constructed based on the prior of the region. The uncalibrable parameter inference module is used to query the state quantile transition matrix with the quantile state of the calibrable parameter as the index, obtain the conditional probability distribution of each uncalibrable parameter in each quantile state, and define the predicted calibration parameter value of each uncalibrable parameter based on the conditional probability distribution. The digital twin simulation and resilience evaluation module is used to combine the measured calibration parameter values ​​and the predicted calibration parameter values ​​to construct a digital twin building for the target building, simulate the structural response under the set seismic loading conditions, and obtain the seismic resilience monitoring results based on the corresponding evaluation of the structural response simulation results.

Images