Tibet dwelling classification vulnerability evaluation method based on earthquake damage data and numerical simulation parameter correction

Abstract

The invention provides a classification vulnerability evaluation method for traditional dwellings in Tibet regions. The method comprises the following steps: step 1, historical earthquake damage statistics; 2, deducing an empirical matrix; step 3, generating seismic oscillation; 4, establishing a typical structure nonlinear value; 5, comparing and correcting data; step 6, establishing a combined vulnerability model; and step seven, verification and uncertainty quantification are carried out. According to the method, a bidirectional correction mechanism of an empirical earthquake damage matrix and a numerical vulnerability curve is fused, and Beta distribution probability deduction and Monte Carlo uncertainty propagation are introduced, so that regional and structural accurate evaluation is realized. The result output comprises a damage probability matrix, a vulnerability curve (including a confidence interval) and risk level suggestions. The method is systematized and repeatable, has regional generalization, is suitable for earthquake disaster risk assessment, disaster reduction planning and seismic reinforcement priority division, and has high engineering and application values.

Description

Technical Field

[0001] This invention belongs to the field of earthquake engineering and earthquake disaster risk assessment, and specifically relates to a building vulnerability modeling method that combines historical earthquake damage data with numerical dynamic analysis. It is particularly suitable for classification vulnerability analysis and model construction of traditional dwellings in plateau areas (such as Tibet) where earthquake damage data is insufficient. Background Technology

[0002] Existing technologies for building vulnerability analysis mainly include empirical statistical methods, seismic damage matrix methods, mathematical fitting methods (such as log-normal distribution and Beta distribution), and numerical simulation methods based on structural dynamics (such as nonlinear time history analysis and incremental dynamic analysis, IDA). Among these, empirical statistical and seismic damage matrix methods rely on historical seismic damage data and can reflect macroscopic damage patterns. However, in areas with insufficient seismic damage data, the limited sample size and intensity coverage make it difficult to accurately describe the damage probability distribution of buildings in high-intensity areas. While numerical simulation methods can characterize structural response characteristics from a mechanical mechanism perspective, their results are greatly affected by modeling assumptions, material parameters, and seismic input characteristics, and require systematic verification and correction with measured seismic damage data.

[0003] In Tibet, my country, traditional dwellings are diverse, including stone-wood, earth-wood, and brick-wood structures, with significant differences in construction techniques, material properties, and connection methods. Meanwhile, the region's complex geological conditions result in earthquakes characterized by strong high-frequency components and pronounced site effects. Due to Tibet's vast size and sparse population, historical earthquake records and damage data are concentrated in low-to-medium intensity areas (VI–VIII), with severely insufficient data for high-intensity areas (IX and above). Therefore, establishing a vulnerability matrix for traditional dwellings using traditional methods based on Tibet's earthquake damage data presents significant challenges, while general vulnerability matrices based on national statistical data exhibit poor applicability and systematic bias in Tibet.

[0004] Existing research still has shortcomings in the following aspects: (1) The spatial and intensity distribution of historical earthquake damage data is uneven and the sample size is limited, making it difficult to support quantitative extrapolation of high-intensity areas; (2) The unique seismic motion spectrum characteristics and site conditions of plateau areas have not been fully considered in vulnerability modeling; (3) There is a lack of a unified mutual calibration and parameter correction mechanism between the vulnerability curve obtained by numerical simulation and the empirical earthquake damage matrix, resulting in insufficient consistency of model results between areas with detailed data and areas with insufficient data, which limits the promotion and application of the method.

[0005] Therefore, there is an urgent need for a systematic technical system that can organically combine empirical earthquake damage data, probability distribution extrapolation methods, vulnerability curve analysis based on seismic ground motion intensity indices (such as PGA, PGV, Sa, etc.), incremental dynamic analysis (IDA), and zonal correction methods based on the level of detail of regional data. By establishing a classification vulnerability assessment model that integrates multi-source information, the accuracy, comparability, and regional applicability of seismic vulnerability analysis for traditional residential buildings in Tibet can be significantly improved. Summary of the Invention

[0006] To overcome the aforementioned deficiencies of existing technologies, this invention provides a method for classifying and assessing the vulnerability of Tibetan dwellings based on earthquake damage data and numerical simulation.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for assessing the vulnerability of dwellings in Tibet based on earthquake damage data and numerical simulation correction, the method comprising the following steps: Step 1: Historical earthquake damage statistics: Collect and standardize earthquake damage survey data of Tibet and surrounding areas since 1990, and statistically analyze the data according to structural type, earthquake intensity and damage level to form an initial empirical earthquake damage matrix; Step 2, Empirical Matrix Deduction: For structural types lacking high-intensity (IX and X) data, a reference standard matrix is ​​used as prior information. By fitting the Beta distribution or log-normal distribution, the probability matrix of damage in high-intensity areas is deduced based on the mean and variance of low-intensity data, resulting in the deduced empirical earthquake damage matrix. Step 3: Earthquake Ground Motion Generation: By comparing and analyzing the regional earthquake ground motion records with the characteristics of earthquake ground motion records in Tibet based on detailed strong earthquake observation data, we will conduct earthquake ground motion analogy and parameter propagation at different intensity levels, establish a method for generating earthquake ground motion records adapted to the characteristics of plateau regions, and provide the uncertainty confidence interval for the generated records. Step 4: Establish nonlinear numerical values ​​for typical structures, input seismic ground motion records (PGA, PGV, Sa) adapted to the characteristics of plateau regions, use the IDA method to calculate the probability of failure under different earthquake intensities, and generate vulnerability curves with seismic ground motion intensity index (IM) as independent variables. Step 5: Data Comparison and Correction: Align the vulnerability curves obtained from numerical simulation with the empirical seismic damage matrix through intensity-PGA mapping, consider the influence of construction age, compare the differences in failure probabilities, and based on the mean and variance of the vulnerability curves obtained from the low-intensity empirical seismic damage matrix and the IDA method, deduce the failure probability matrix of the high-intensity area through Beta distribution or log-normal distribution fitting method to obtain the deduced empirical seismic damage matrix; Step 6: Establishment of Joint Vulnerability Model: Integrating various structural types, intensity levels, and regional characteristics, construct a classification vulnerability matrix and vulnerability curve set with uncertainty parameters (median and dispersion coefficient β), and output standardized data format for use in earthquake risk assessment and decision support; Step 7, Validation and Uncertainty Quantification: Through cross-validation, sensitivity analysis and Monte Carlo simulation, quantify the uncertainty of each parameter, calculate the confidence interval of the model prediction results and the recommended safety factor.

[0008] As a further improvement to the technical solution of the present invention, the probability distribution model used in step two is preferably a Beta distribution, and its shape parameters are determined by maximum likelihood estimation (MLE) or moment matching method.

[0009] As a further improvement to the technical solution of the present invention, step four adopts a nonlinear finite element (FEM) or finite-discrete element (FDEM) hybrid modeling method to simulate the seismic response behavior of stone-wood, earth-wood and brick-wood structures.

[0010] As a further improvement to the technical solution of the present invention, the correction of the numerical simulation curve in step five adopts a posterior correction model based on Bayesian update, in which the empirical matrix is ​​used as the likelihood function and the numerical vulnerability curve is used as the prior distribution to obtain the optimal correction result.

[0011] As a further improvement to the technical solution of the present invention, step five determines the transmission range of the correction factor through a regional similarity evaluation mechanism. The similarity is calculated based on material type, construction process and site category, and the value range is 0 to 1.

[0012] As a further improvement to the technical solution of the present invention, in step six, the fragility curve is fitted using a log-normal distribution.

[0013] As a further improvement to the technical solution of the present invention, step seven adopts a method combining Monte Carlo random sampling and confidence interval estimation to perform joint propagation analysis on parameter error, model error and ground motion uncertainty, and outputs a 5%–95% confidence interval range.

[0014] A method for assessing the vulnerability of dwellings in Tibet based on seismic damage data and numerical simulation correction, characterized in that the output of the method includes: (1) Damage probability matrix classified by intensity or PGA; (2) Vulnerability curves for each structural type (including median and confidence interval); (3) Uncertainty quantification report and explanation of the applicable scope of the model.

[0015] A method for assessing the vulnerability of residential buildings in Tibet based on earthquake damage data and numerical simulation correction is characterized by the ability of the method to automatically call the corresponding modules based on the input layer information (administrative region, latitude and longitude, structural type, year of construction, number of floors, site category, seismic motion characteristics and data detail identifier) ​​to achieve classified vulnerability analysis.

[0016] The beneficial effects of this invention are: This invention establishes a two-way correction mechanism between empirical earthquake damage matrices and numerical vulnerability curves by systematically integrating historical earthquake damage data with numerical simulation analysis methods, achieving synergistic unification of empirical data and physical models. This method fully considers the characteristics of strong high-frequency components and complex site conditions in the Tibetan Plateau region, correcting the spectral characteristics of input earthquake ground motions to make vulnerability assessment results more consistent with regional realities. Addressing the problem of insufficient high-intensity data, this invention employs a probabilistic extrapolation and regional analogy correction method based on Beta distribution, significantly improving the reliability of earthquake damage prediction in data-scarce areas. By introducing an age correction coefficient, the impact of differences in construction techniques and material properties of residential buildings from different eras on seismic performance can be quantitatively reflected. Finally, the method outputs a vulnerability matrix and curves with confidence intervals, which can be embedded in earthquake risk assessment and emergency decision-making systems, providing a scientific basis for earthquake damage prediction, disaster assessment, and priority allocation for residential building reinforcement. It has high engineering applicability, regional scalability, and academic reference value. Attached Figure Description

[0017] Figure 1 This is a flowchart of the Tibetan residential building classification and vulnerability assessment method of the present invention.

[0018] Figure 2 This is a seismic vulnerability curve used as a verification case for the present invention. Detailed Implementation

[0019] The implementation process, key algorithms, and parameter settings of the present invention are described in detail below using examples. For ease of engineering application, this embodiment is explained in seven steps (e.g., Figure 1 As shown in the figure, the data processing and analysis requirements for each step are given.

[0020] 1. Data collection and standardization (1) Data sources: compilations of earthquake damage surveys by national and provincial earthquake bureaus, field survey forms, local disaster damage assessment reports, academic literature and historical archives.

[0021] (2) Record fields (required): event number, date of occurrence, magnitude, intensity, observation station PGA / PGV, administrative unit of the disaster area, building structure type (stone wood / earth wood / brick wood, etc.), building age, number of floors, building area, damage level.

[0022] (3) Field cleaning: The damage level is uniformly classified as level three (basically intact / damaged / ruined) or level five (basically completed / minor damage / moderate damage / severe damage / collapsed); duplicate and obviously erroneous records are removed, and missing values ​​are marked.

[0023] 2. Spatial / temporal partitioning Spatial partitioning is performed by province / autonomous prefecture / county; and segmentation is performed by time window (e.g., before 1989, 1990–2000, 2001–2010, 2011–present) to assess the differences in seismic performance in different periods.

[0024] Step A: Historical earthquake damage data statistics; For each structural type and intensity zone, the frequency of damage levels (sample ratio) is statistically analyzed to form a standardized empirical earthquake damage matrix with intensity as the row and damage level as the column. Output: Empirical earthquake damage matrix table (by structural type and time / spatial partition).

[0025] For cases with limited sample size, Wilson confidence intervals or Bayesian Beta posterior can be used for smooth estimation to avoid extreme values ​​of zero probability or unit probability; events with similar meanings can be merged or assigned to the closest intensity level according to PGA mapping.

[0026] Step B: Empirical earthquake damage matrix derivation; A standard matrix with similar seismic resistance capacity is selected as the prior (e.g., the D-type matrix from Yin Zhiqian 1996). The "seismic damage index" corresponding to each damage level is treated as a continuous random variable, and a frequency histogram is constructed (segmented with an accuracy of 0.01). A Beta distribution or log-normal distribution is used for fitting, and the shape parameters (α, β) are determined through maximum likelihood estimation (MLE) or moment matching. Using the mean and variance of low-intensity data (VI–VIII) and referring to high-intensity data from the standard matrix, the probability of damage at the target intensity is extrapolated through interpolation / extrapolation or Bayesian updates. The probability density function is integrated to obtain the probability of each damage level, forming the extrapolated seismic damage matrix. The matrices and confidence intervals for stone and wood, earth and wood, and brick and wood at high intensities such as IX and X are output.

[0027] Step C: Numerical simulation to compare and analyze the differences in seismic performance of buildings from different eras; Representative models (such as traditional stone and wood type, improved type, modern civil engineering type, etc.) are determined based on architectural drawings and site surveys, and different material parameters and connection strengths are assigned to them according to different eras / construction techniques. Nonlinear beam-column elements, masonry finite element methods, or finite-discrete hybrid models (FEM / FDEM) are used to simulate wall failure modes. Representative plateau ground motion records or synthetic spectrum seismic records are input for nonlinear time history analysis or IDA to obtain indicators such as peak displacement, maximum shear force, energy dissipation, failure mode, and exceedance probability of failure level. By comparing the vulnerability curves of models from different eras, the exceedance probability ratio under the same IM is calculated, and a time correction coefficient k is defined. A table of differences in vulnerability curves from different eras and a table of time correction coefficients are output.

[0028] Step D: Establish vulnerability curves by considering incremental dynamic analysis of plateau seismic motion characteristics; Select or synthesize representative seismic records from the plateau region to ensure that the spectral width covers high-frequency energy; scale the records to cover the target PGA range, and perform IDA at each level until the structure reaches the failure or ultra-large deformation threshold; obtain structural response indicators (such as the maximum inter-story drift angle), and calculate the exceedance probability of each failure level under each PGA level; use a log-normal distribution or point-by-point Bayesian method to fit and generate a smooth vulnerability curve, and output the median, scattering parameter (β), and confidence interval.

[0029] Step E: Comparison and correction of numerical simulation curves with empirical matrices; If the empirical matrix is ​​reliable in the data-rich region, local weighted correction can be performed on the numerical simulation curve (the correction function R can be proportional, power law, or polynomial least squares). If there is a systematic bias in the numerical model, Bayesian update can be used, taking the numerical curve as the prior and the empirical matrix as the likelihood, to obtain the posterior curve. Output the corrected numerical fragility curve and the regional correction coefficient table, and explain the applicable scope.

[0030] Step F: Refine the empirical earthquake damage matrix; The correction factor obtained by comparing the correction matrix and numerical curve in the data-rich area is propagated to the same structure or similar site in the data-deficient area (based on the similarity of materials, structure, and site conditions); the mapping rule is defined and the similarity is quantified (0-1); the derivation matrix in step B is used as the initial value, and then analogy correction is applied to generate the final empirical matrix; the improved empirical earthquake damage matrix (by zone / structure / intensity) and uncertainty interval are output.

[0031] Step G: Establish a classification vulnerability assessment model for Tibetan dwellings; Earthquake disaster loss compilation data was used, combined with records from the Tibet region for seismic motion correction considering regional characteristics; the Beta distribution was preferred for fitting the probability distribution of damage levels, and the log-normal distribution was preferred for fitting the vulnerability curve; the IDA record set was recommended to be ≥21 records, covering the spectral width, with a scaling factor PGA of 0.1g–3.0g; log-space linear regression or weighted least squares fitting was used; Monte Carlo simulation was used for joint uncertainty propagation, outputting 5%–95% confidence intervals.

[0032] Validation and Application Examples The structural damage ratios of residential buildings in Tibet from 1993 to the present were obtained. Based on the relationship between ground motion and damage ratio in the China Seismic Intensity Scale (CB / T17742-2020), a log-normal distribution was used to fit the vulnerability curves of residential buildings in Tibet. The seismic vulnerability curves of stone-timber, earth-timber, and brick-timber structures in Tibet under different damage levels are shown below. Figure 2 As shown in the figure. The results demonstrate that the method of this invention can accurately predict the seismic vulnerability of residential buildings of various structural types in Tibet, verifying the effectiveness and regional applicability of the method.

Claims

1. A method for assessing the vulnerability of residential buildings in Tibet based on seismic damage data and numerical simulation correction, characterized in that, The method includes the following steps: Step 1: Historical earthquake damage statistics: Collect and standardize earthquake damage survey data of Tibet and surrounding areas since 1990, and statistically analyze the data according to structural type, earthquake intensity and damage level to form an initial empirical earthquake damage matrix; Step 2, Empirical Matrix Deduction: For structural types lacking high-intensity (IX and X) data, a reference standard matrix is ​​used as prior information. By fitting the Beta distribution or log-normal distribution, the probability matrix of damage in high-intensity areas is deduced based on the mean and variance of low-intensity data, resulting in the deduced empirical earthquake damage matrix. Step 3: Earthquake Ground Motion Generation: By comparing and analyzing the regional earthquake ground motion records with the characteristics of earthquake ground motion records in Tibet based on detailed strong earthquake observation data, we will conduct earthquake ground motion analogy and parameter propagation at different intensity levels, establish a method for generating earthquake ground motion records adapted to the characteristics of plateau regions, and provide the uncertainty confidence interval for the generated records. Step 4: Establish nonlinear numerical values ​​for typical structures, input seismic ground motion records (PGA, PGV, Sa) adapted to the characteristics of plateau regions, use the IDA method to calculate the probability of failure under different earthquake intensities, and generate vulnerability curves with seismic ground motion intensity index (IM) as independent variables. Step 5: Data Comparison and Correction: Align the vulnerability curves obtained from numerical simulation with the empirical seismic damage matrix through intensity-PGA mapping, consider the influence of construction age, compare the differences in failure probabilities, and based on the mean and variance of the vulnerability curves obtained from the low-intensity empirical seismic damage matrix and the IDA method, deduce the failure probability matrix of the high-intensity area through Beta distribution or log-normal distribution fitting method to obtain the deduced empirical seismic damage matrix; Step 6: Establishment of Joint Vulnerability Model: Integrating various structural types, intensity levels, and regional characteristics, construct a classification vulnerability matrix and vulnerability curve set with uncertainty parameters (median and dispersion coefficient β), and output standardized data format for use in earthquake risk assessment and decision support; Step 7, Validation and Uncertainty Quantification: Through cross-validation, sensitivity analysis and Monte Carlo simulation, quantify the uncertainty of each parameter, calculate the confidence interval of the model prediction results and the recommended safety factor.

2. The method for assessing the vulnerability of dwellings in Tibet based on earthquake damage data and numerical simulation correction as described in claim 1, characterized in that, The probability distribution model used in step two is preferably a Beta distribution, and its shape parameters are determined by maximum likelihood estimation (MLE) or moment matching method.

3. The method for assessing the vulnerability of dwellings in Tibet based on earthquake damage data and numerical simulation correction as described in claim 1, characterized in that, Step four employs a hybrid modeling method of nonlinear finite element (FEM) or finite-discrete element (FDEM) to simulate the seismic response behavior of stone-wood, earth-wood, and brick-wood structures.

4. The method for assessing the vulnerability of dwellings in Tibet based on earthquake damage data and numerical simulation correction as described in claim 1, characterized in that, In step five, the correction of the numerical simulation curve adopts a posterior correction model based on Bayesian update, in which the empirical matrix is ​​used as the likelihood function and the numerical vulnerability curve is used as the prior distribution to obtain the optimal correction result.

5. The method for assessing the vulnerability of dwellings in Tibet based on earthquake damage data and numerical simulation correction as described in claim 1, characterized in that, Step five determines the range of correction factor propagation through a regional similarity evaluation mechanism. The similarity is calculated based on material type, construction technology and site category, and the value range is 0 to 1.

6. The method for assessing the vulnerability of dwellings in Tibet based on earthquake damage data and numerical simulation correction as described in claim 1, characterized in that, In step six, the fragility curve is fitted using a log-normal distribution.

7. The method for assessing the vulnerability of dwellings in Tibet based on earthquake damage data and numerical simulation correction as described in claim 1, characterized in that, Step seven employs a combination of Monte Carlo random sampling and confidence interval estimation to perform joint propagation analysis on parameter errors, model errors, and ground motion uncertainties, outputting a 5%–95% confidence interval range.

8. A method for assessing the vulnerability of dwellings in Tibet based on seismic damage data and numerical simulation correction, as described in any one of claims 1 to 7, characterized in that... The output of the method includes: (1) Damage probability matrix classified by intensity or PGA; (2) Vulnerability curves for each structural type (including median and confidence interval); (3) Uncertainty quantification report and explanation of the applicable scope of the model.

9. A method for assessing the vulnerability of dwellings in Tibet based on seismic damage data and numerical simulation correction, as described in any one of claims 1 to 8, characterized in that... The method can automatically call the corresponding modules based on the input layer information (administrative region, latitude and longitude, structure type, year of construction, number of floors, site category, seismic characteristics and data detail identifier) ​​to achieve classification vulnerability analysis.

Images