A method and system for evaluating seismic damage of a sedimentary basin building group

Abstract

The application discloses a sedimentary basin building group earthquake disaster evaluation method and an evaluation system, belongs to the technical field of engineering disaster prevention, and is characterized in that the sedimentary basin building group earthquake disaster evaluation method comprises the following steps: S1, determining a target site and characteristic input parameters; S2, establishing a three-dimensional sedimentary basin model and obtaining a seismic response; S3, establishing an improved data set; S4, optimizing an artificial neural network; S5, obtaining samples; S6, determining seismic motion parameters; S7, prediction and evaluation. The application improves the calculation efficiency of building the data set and evaluating the influence of site medium parameter uncertainty on basin seismic motion by means of a fast multipole algorithm and a surrogate model in sequence, can give statistical moment and probability information of peak acceleration of the basin ground surface at any position, and is applied to rapid prediction and evaluation of building group earthquake disaster, and serves a resilient city.

Description

Technical Field

[0001] This invention belongs to the field of engineering disaster prevention technology, specifically relating to a method and system for assessing earthquake damage to building complexes in sedimentary basins. Background Technology

[0002] In determining seismic motion parameters, designing and analyzing engineering structures in complex sites, the scientific determination of seismic motion parameters is crucial. Many cities in my country, such as Beijing, Urumqi, and Lanzhou, are located in sedimentary basins. Due to the significant differences between the soil and rock properties within the basin and the outer bedrock, seismic wave scattering occurs, significantly impacting the temporal, spatial, and intensity distribution of ground motions. This manifests as a center-focusing effect, an edge-amplification effect, and long-period characteristics. Multiple earthquake damage studies and theoretical research have shown that basin effects exacerbate building damage. Limited by soil testing conditions, site medium parameters often exhibit uncertainty, and this uncertainty is transmitted during seismic wave propagation, leading to uncertainty in building damage. Therefore, conducting seismic damage assessments of building complexes in sedimentary basins that consider the uncertainties of the site medium is of great significance.

[0003] The prerequisite for predicting and assessing seismic damage to sedimentary basin ensembles is the scientific determination of seismic motion parameters that take into account basin effects and site medium uncertainties. Domestic and international scholars have studied the surface seismic response and characteristics of sedimentary basins using analytical and numerical methods. Analytical methods are suitable for basins with simple geometry and material conditions; for sedimentary basins with complex geometries or wave velocity structures, numerical simulation methods based on three-dimensional wave theory have wider applicability. Existing research results show that both incident wave characteristics and site medium properties significantly affect the seismic response of sedimentary basins. However, classical regression techniques struggle to quantify the highly nonlinear relationship and multi-factor characteristics between incident waves, site medium properties, and the seismic response of sedimentary basins. Furthermore, numerical simulation methods for analyzing the uncertainty of three-dimensional sedimentary basin seismic motion often require high computational costs, hindering rapid regional risk assessment. Therefore, a method for solving the three-dimensional seismic response of sedimentary basins that can simultaneously consider site effects and site medium uncertainties is urgently needed. Summary of the Invention

[0004] This invention addresses the technical problems existing in prior art by providing a method and system for assessing seismic damage to sedimentary basin building complexes. To improve the computational efficiency of conventional methods, a seismic damage assessment scheme for sedimentary basin building complexes considering the uncertainties of the site medium is proposed. This scheme can automatically extract and concretize the abstract features of complex site effects, construct a data-driven mapping relationship, and parameterize this mapping relationship using a neural network. While ensuring accuracy, it enables efficient response simulation of the target problem.

[0005] The first objective of this invention is to provide a method for assessing seismic damage to sedimentary basin ensembles, comprising:

[0006] S1. Determine the target site and feature input parameters:

[0007] Sedimentary basins were selected as target sites. The seismic motion frequency bands, site medium property parameters and their value ranges in actual earthquakes were determined and used as feature input parameters in the basic dataset of artificial neural networks.

[0008] S2. Establish a three-dimensional sedimentary basin model and obtain the seismic response:

[0009] A three-dimensional sedimentary basin model is established, and the feature input parameters are specified. Each feature input parameter is randomly selected within a given value range. The seismic response of the three-dimensional sedimentary basin model under any set of random numbers is solved using the fast multipole boundary element method. The seismic responses under multiple sets of random numbers constitute an elementary dataset.

[0010] S3. Establish an improved dataset:

[0011] Based on the elementary dataset, an improved dataset is established, which uses the relative position of the sedimentary basin surface as the feature input parameter, and the incident frequency of bedrock seismic waves, the material properties of the medium in the sedimentary basin, and the relative position of the sedimentary basin surface as the feature input parameter, and the displacement amplification factor (DAF) at each position of the sedimentary basin surface as the feature output parameter.

[0012] S4. Optimize the artificial neural network;

[0013] S5. Obtaining Samples:

[0014] A real-world case model for solving the target problem is established. The trained neural network is used as a proxy model to analyze the uncertainty of seismic response in three-dimensional sedimentary basins. The basin seismic response corresponding to each sample in the target problem is solved, and the Monte Carlo method for analyzing uncertainty problems is quickly provided with the required samples.

[0015] S6. Determination of seismic motion parameters:

[0016] Based on the actual sedimentary basin conditions, uncertain parameters of the site medium, their probability distributions, and value ranges are defined. It is determined whether there are samples in the basic dataset that overlap with the uncertain features. If so, the results are extracted from the corresponding improved dataset. If not, the surrogate model is called to calculate the uncertain feature combinations sequentially to obtain the statistical moments of the surface displacement amplification factor, velocity, and acceleration amplitude of the sedimentary basin. The peak acceleration statistical moments at any location on the surface of the sedimentary basin and the peak acceleration corresponding to any confidence level are obtained using the inverse fast Fourier transform. The multi-dimensional results are used to conduct a quantitative assessment of the impact of uncertainties in site medium parameters on three-dimensional sedimentary basin ground motion.

[0017] S7. Forecasting and Assessment:

[0018] Based on actual needs, select peak ground acceleration of sedimentary basin surface ground corresponding to different confidence levels, and combine with local building seismic vulnerability curve database or vulnerability analysis to carry out seismic damage prediction and assessment of sedimentary basin building clusters considering the uncertainty of site medium parameters.

[0019] Preferably, in S2, the characteristic input parameters are specified as the incident frequency of bedrock seismic waves, the ratio of shear wave velocities inside and outside the sedimentary basin, the damping ratio, and the Poisson's ratio.

[0020] Preferably, S4 specifically involves: selecting an artificial neural network structure, dividing the improved dataset to obtain a training set, a test set, and a validation set, training the artificial neural network based on the training set, introducing a differential evolution-particle swarm optimization algorithm to optimize the initial weights and thresholds during the training process, selecting parameters and testing the network accuracy of the artificial neural network based on the test set and the validation set, and using the artificial neural network with the best test performance as a data-driven surrogate model for solving the target problem.

[0021] A second objective of this invention is to provide a seismic damage assessment system for sedimentary basin building complexes, comprising:

[0022] Data acquisition module: Determines the target site and feature input parameters, selects sedimentary basins as the target site, determines the seismic frequency bands of sedimentary basins in actual earthquakes, the site medium property parameters and their value ranges of sedimentary basins, and uses them as feature input parameters in the basic dataset of artificial neural networks;

[0023] Model building module: Build a three-dimensional sedimentary basin model and obtain the seismic response. Build a three-dimensional sedimentary basin model and specify the feature input parameters. Take random numbers for each feature input parameter within a given value range. Use the fast multipole boundary element method to solve the seismic response of the three-dimensional sedimentary basin model under any set of random numbers. The seismic responses under multiple sets of random numbers form an elementary dataset.

[0024] Dataset Improvement Module: An improved dataset is established. Based on the elementary dataset, the relative position of the sedimentary basin surface is also used as a feature input parameter. The improved dataset uses the incident frequency of bedrock seismic waves, the material properties of the medium in the sedimentary basin, and the relative position of the sedimentary basin surface as feature input parameters, and the displacement amplification factor (DAF) at each position of the sedimentary basin surface as feature output parameters.

[0025] Optimization module: Optimizes artificial neural networks;

[0026] Sample acquisition module: Acquires samples, establishes a real-world case model to solve the target problem, uses the trained neural network as a proxy model to analyze the uncertainty of seismic response in three-dimensional sedimentary basins, solves the basin seismic response corresponding to each sample in the target problem, and quickly provides the required samples for Monte Carlo method to analyze uncertainty problems;

[0027] Parameter Determination Module: Seismic motion parameter determination. Based on the actual sedimentary basin conditions, it sets uncertain parameters of the site medium, their probability distributions, and value ranges. It determines whether there are samples in the basic dataset that overlap with the uncertain features. If so, it extracts the results from the corresponding improved dataset. If not, it calls the surrogate model to calculate the uncertain feature combinations sequentially, obtaining the statistical moments of the sedimentary basin surface displacement amplification factor, velocity, and acceleration amplitude. It uses the inverse fast Fourier transform to obtain the peak acceleration statistical moments at any location on the sedimentary basin surface and the peak acceleration corresponding to any confidence level. It uses multi-dimensional results to conduct a quantitative assessment of the impact of site medium parameter uncertainties on three-dimensional sedimentary basin seismic motion.

[0028] Execution module: Prediction and assessment. Based on actual needs, select peak ground acceleration of sedimentary basin surface ground corresponding to different confidence levels, and combine local building seismic vulnerability curve database or vulnerability analysis to carry out seismic damage prediction and assessment of sedimentary basin building clusters considering the uncertainty of site medium parameters.

[0029] Preferably, in the model building module, the feature input parameters are specified as the incident frequency of bedrock seismic waves, the ratio of shear wave velocities inside and outside the sedimentary basin, the damping ratio, and the Poisson's ratio.

[0030] Preferably, the optimization process of the optimization module is as follows: selecting an artificial neural network structure, dividing the improved dataset to obtain a training set, a test set, and a validation set, training the artificial neural network based on the training set, introducing differential evolution-particle swarm optimization algorithm to optimize the initial weights and thresholds during the training process, selecting parameters and testing the network accuracy of the artificial neural network based on the test set and validation set, and using the artificial neural network with the best test performance as a data-driven surrogate model for solving the target problem.

[0031] A third objective of this invention is to provide an information data processing terminal for implementing the aforementioned method for assessing earthquake damage to sedimentary basin building complexes.

[0032] A fourth objective of this invention is to provide a computer-readable storage medium including instructions that, when executed on a computer, cause the computer to perform the above-described method for assessing seismic damage to sedimentary basin ensembles.

[0033] The advantages and positive effects of this invention are:

[0034] 1) The technical solution proposed in this invention can accurately simulate the seismic response of sedimentary basins under the coupled effects of various uncertain factors in the site medium, and can balance the calculation accuracy and calculation efficiency. In order to accurately describe the influence of uncertain parameters of the site medium on the seismic response of sedimentary basins, the Monte Carlo method is generally used, but it has a large number of calculation samples and the calculation cost of a single sample is high. Using an artificial neural network surrogate model to replace numerical simulation can greatly improve the calculation efficiency of a single sample, thereby achieving efficient assessment of uncertainty.

[0035] 2) The application of this invention can reduce the assumption of artificial conditions and significantly reduce the difficulty of analyzing the seismic response of complex sites under the action of uncertain factors, making it convenient for structural engineers to select engineering sites and determine engineering seismic parameters. This invention has wide applicability, and other types of complex sites can also use this method to construct proxy models, carry out the quantification of seismic uncertainty in complex sites and the assessment of seismic damage to building groups. Attached Figure Description

[0036] Figure 1 This is a flowchart of an embodiment of the present invention;

[0037] Figure 2 The flowchart is shown below for the ADE-PSO algorithm.

[0038] Figure 3 A schematic diagram of the structure of an artificial neural network

[0039] Figure 4 This is a schematic diagram of a sedimentary basin.

[0040] Figure 5 This is a comparison chart of the neural network's predicted values ​​and the actual values.

[0041] Figure 6 This is a performance evaluation graph for neural networks;

[0042] Figure 7 The time-domain response statistical moments and probability information diagrams are provided for the surrogate model;

[0043] Figure 8 This is a failure probability diagram corresponding to different damage levels of a building complex when the peak ground acceleration is used as the seismic intensity index at a confidence level of 85%. Detailed Implementation

[0044] To further understand the invention's content, features, and effects, the following embodiments are provided, and detailed descriptions are given below in conjunction with the accompanying drawings:

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the technical solutions of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] Artificial intelligence technologies are increasingly being used to solve earthquake engineering problems, with artificial neural networks (ANNs) being the most widely applied. An ANN is a multi-layer feedforward neural network algorithm that combines backpropagation with a neural network structure. It does not require pre-assumed mapping relationships, making it suitable for constructing highly nonlinear mappings. A well-trained ANN can serve as a surrogate model for solving related problems. Therefore, the basic idea of ​​this invention is to construct the dataset required to build an ANN surrogate model by calculating a small number of samples. A differential particle swarm optimization algorithm is introduced to initially optimize the weights and thresholds. After training and testing to obtain a surrogate model with good accuracy, the Monte Carlo method is used to assess the uncertainty of ground motion in three-dimensional sedimentary basins. Single-sample results are obtained through the surrogate model or directly extracted from existing results in the dataset, eliminating the need for numerical simulation. By improving the efficiency of single-sample calculations, the computational cost of the Monte Carlo method is reduced, achieving the quantification of ground motion uncertainty in three-dimensional sedimentary basins and applying it to the rapid prediction and assessment of earthquake damage to building clusters. To quickly establish the dataset, the fast multipole boundary element method is used to solve the seismic response of three-dimensional sedimentary basins under deterministic parameter conditions.

[0047] Please see Figures 1 to 8 .

[0048] A method for assessing seismic damage to sedimentary basin ensembles includes the following steps:

[0049] Step 1: Establish a physical driving model for solving the target problem.

[0050] To determine the target site, given the wide distribution of sedimentary basins and the need to consider their geological and geomorphological characteristics in urban planning and design, this application selects a semi-ellipsoidal sedimentary basin as the target site. A three-dimensional model is used to accurately account for the scattering effect of seismic waves on the sedimentary basin.

[0051] Step 2: Solve the response of the three-dimensional sedimentary basin under seismic excitation using numerical simulation.

[0052] Figure 4This invention presents a numerical model of a sedimentary basin built on a programming platform. The Fast Multipole Boundary Element Method (FME) is used to simulate the response of a sedimentary basin to bedrock seismic waves. The FME significantly reduces computational load and storage space, improving the efficiency of building a foundational dataset for artificial neural networks. First, the FME preprocesses the potential function. Key steps include: 1) introducing cell nodes in a hierarchical tree structure to convert direct interactions between elements into indirect interactions with these cell nodes as intermediate points; and 2) expanding the Green's function using Taylor series, essentially a rationalization of complex functions. Next, the sedimentary basin surface, bedrock surface, and bedrock-sedimentary interface are divided into elements. Then, assuming the sedimentary basin does not exist, the free-field response is obtained. The seismic excitation is then equivalent to a virtual load through the Green's function and applied to the discrete elements. The scattered field response is solved using boundary conditions and the Green's function. Finally, the free-field and scattered field responses are combined to obtain the overall site response. In this invention, the overall site effect is defined as the frequency-domain displacement amplitude. Expanding the obtained frequency-domain displacement amplitude using fundamental physical relationships yields the frequency-domain velocity and acceleration amplitudes. The fast multipole boundary element method algorithm can be directly implemented based on an open-source programming platform, and the physical model of a three-dimensional sedimentary basin can also be directly established through a programming language, which greatly reduces the operational difficulties of various complex software.

[0053] Step 3: Determine the uncertain parameters and their range of values.

[0054] After comprehensively considering existing databases, relevant literature, and national standards, this invention identifies the uncertainties affecting the seismic response of sedimentary basins and their value ranges. It also clarifies the model type, modeling method, site medium and geometric properties, and boundary conditions. This invention selects the ratio of internal to external shear wave velocities, Poisson's ratio, and damping ratio of the sedimentary basin as uncertain parameters, representing proxy conditions characterizing the uncertainty of site medium properties. The incident wave frequency is also included as an uncertain parameter, with a given value range.

[0055] Step 4: Establish the basic dataset of "Uncertain parameters - seismic response of sedimentary basins".

[0056] The random seed command based on the Matlab programming platform generates corresponding random data sets for the four random variables determined in step 3. Each data set represents a random variable, and the data within each set must meet the set reference range and the dimensions of the four data sets must be consistent. The four data sets form a basic dataset, which is input into the physical model established in step 2. The three-dimensional sedimentary basin seismic effect is calculated sequentially under the conditions of each dataset using the fast multipole boundary element method. After the simulation is completed, the frequency domain displacement amplitude of all basin surface points is extracted from the simulation results. To reduce the number of output parameters of the artificial neural network, the Cartesian coordinates of the basin surface location are also used as input feature parameters for constructing the surrogate model. The sorting command is used to orderly combine the basic dataset, the frequency domain displacement amplitude of any basin surface point, and the location coordinates to form an improved dataset. Each sample in the improved dataset represents the frequency domain displacement amplitude of a certain surface point in the sedimentary basin under the condition that the four uncertain parameters are constant.

[0057] Step 5: Establish a proxy model for analyzing uncertainties in three-dimensional sedimentary basins.

[0058] The improved dataset obtained in step 4 is randomly divided into a training set and a test set. The artificial neural network is trained using the training set, and the performance of the artificial neural network is tested using the test set. The ratio of the training set to the test set is 8:2.

[0059] Figure 2 This document describes the flow of the Differential Evolution-Particle Swarm Optimization (DE-PSO) algorithm. ADE represents Differential Evolution, PSO represents Particle Swarm Optimization, and ANN represents Artificial Neural Network. This algorithm optimizes the weights and thresholds of the input and hidden layers in an artificial neural network. The DE-PSO algorithm randomly generates a population of individuals with corresponding weights and thresholds. In this invention, the weights and thresholds are used as individuals in the population, i.e., the search particles of the DE-PSO algorithm. The parameters of the DE-PSO algorithm are initialized, including the convergence threshold, population size, inertia weight, maximum number of iterations, maximum number of particles, crossover rate, mutation rate, and upper and lower bounds for position.

[0060] Train the artificial neural network and calculate the fitness of each particle. Compare the fitness values ​​to select the locally optimal and globally optimal particles in the particle population. Update the particle velocity and position, encode the new population, and perform selection, genetic crossover, and mutation operations on the population. Determine if the convergence condition is met. If the condition is met, obtain the optimal weights and thresholds, assign them to the artificial neural network, and train it. If the condition is not met, increase the number of iterations and repeat the relevant steps of the optimization algorithm until the convergence condition is met. Obtain the required weights and thresholds and assign them to the artificial neural network for training.

[0061] Configure the network structure parameters of the artificial neural network. For example... Figure 3 As shown, the artificial neural network of this invention adopts a four-layer network structure, including one input layer, two hidden layers, and one output layer. The structural parameters of the artificial neural network include the number of nodes in the input layer, the number of nodes in the two hidden layers, the number of nodes in the output layer, the activation function of each layer, the training function, the learning rate, the set convergence error, the number of training iterations, and the maximum number of failures. The number of nodes in the input layer and the output layer are determined based on the research problem; the input layer has 6 nodes, and the output layer has 1 node. The number of nodes in the two hidden layers is determined using a trial-and-error method. First, a large number of experiments are conducted to roughly determine the range of nodes in the two hidden layers. Then, the neural network within this range is trained using the same training samples. The neural network with the smallest error on the test set is selected as the final surrogate model. The number of nodes in the two hidden layers are 38 and 32, respectively. The activation function of each layer is the tanh function. Gradient descent is used during neural network training for adaptive error adjustment. The training function is the training dm function, and the evaluation functions are MSE and Loss functions. Regularization techniques are added during training to reduce overfitting. Other relevant parameters are determined through extensive experimental methods. A well-trained neural network can serve as a surrogate model for analyzing uncertainties in three-dimensional sedimentary basins. The data to be predicted is first preprocessed by normalization, and then the surrogate model is used to simulate and predict the normalized data to obtain the frequency domain displacement amplitude of each basin surface point under this condition.

[0062] Step 6: Evaluation of the accuracy and applicability of artificial neural networks.

[0063] To fully evaluate the generalization ability of artificial neural networks in solving problems, this invention employs three metrics. Residuals and root mean square error are calculated as dimensionality metrics to assess the deviation between predicted and actual values ​​for individual samples and the test set. The model is also evaluated using the dimensionless coefficient of determination (CCD), ranging from [0,1]. Its purpose is to measure the degree to which the independent variable explains the dependent variable; the closer the CCD is to 1, the higher the applicability of the surrogate model and the more accurately it captures the complex mapping relationship between input and output. Figure 5 The comparison between the prediction results and the expected results given by the artificial neural network is presented. RMSE represents the root mean square error. It can be seen that the comparison between the two results is good within the sample interval under discussion, and the root mean square error is small. Figure 6 The determination coefficients of the artificial neural network on the training and test sets are given. The determination coefficients of both datasets are greater than 0.96, indicating that the surrogate model established in this invention is an excellent model with strong generalization ability to the target problem.

[0064] Step 7: Quantify the uncertainty of ground motion in the three-dimensional sedimentary basin based on the surrogate model.

[0065] The solution to uncertainty problems is often found using the classic Monte Carlo method. The algorithm's development emphasizes the randomness of data distribution and leverages probability and statistics theories. The Monte Carlo method requires calculating a large number of uncertainty samples, using statistical analysis to determine the moment response and probability information of the target problem. The Monte Carlo method demands a large number of samples, which can reach significant amounts under seismic wave excitation at different frequencies. Well-trained artificial neural network surrogate models can be used to analyze uncertainties in three-dimensional sedimentary basins. Their advantage lies in significantly improving computational efficiency compared to traditional numerical simulation methods.

[0066] After training, the artificial neural network can be directly saved using a programming language and called directly on the relevant platform when needed, without further training. After being called, the surrogate model is applied to solve practical problems. Based on the actual sedimentary basin conditions, the range of values ​​for quantitative uncertainties (incident wave frequency, internal / external shear wave velocity ratio, Poisson's ratio, damping) is determined. First, it is determined whether there are samples in the basic dataset that overlap with the uncertainties. If so, the results are directly extracted from the corresponding improved dataset; if not, the neural network is called to calculate the combination of uncertainties sequentially. In this invention, the output parameters of the surrogate model are the frequency domain displacement amplitude of the basin surface points, which, after calculation using the first and second derivatives, yields the frequency domain velocity and acceleration amplitudes. The statistical moments in the frequency domain are then used to obtain the acceleration time history and peak acceleration statistical moments through inverse Fourier transform. Figure 7 Three typical time-domain statistical moment results are presented, demonstrating that the surrogate model established in this study can efficiently determine the ground motion parameters of sedimentary basins under the uncertain coupling effects of bedrock seismic waves and medium parameters using multi-dimensional indices.

[0067] Step 8: Conduct seismic damage assessment of building clusters by combining local building structure seismic vulnerability database.

[0068] Figure 8 As shown, the required confidence level of ground motion parameters is determined according to actual needs. Peak ground acceleration is used as the ground motion intensity parameter. Combined with the existing local building structure vulnerability database, the failure probability corresponding to different damage levels of regional building groups is obtained, which is used for pre-earthquake prediction of unfavorable areas or buildings and for rapid post-earthquake assessment.

[0069] In summary, this invention establishes a physical driving model to solve the target problem and uses numerical simulation to solve the seismic response of a three-dimensional sedimentary basin under a small number of deterministic parameters. It also establishes a basic dataset of "uncertain parameters - seismic response of sedimentary basins," setting the basin surface coordinates as input feature parameters to create an improved dataset. Based on this improved dataset, an artificial neural network is used to establish a high-precision surrogate model for analyzing the uncertainty problem of three-dimensional sedimentary basins. Then, based on the surrogate model, the uncertainty of ground motion in three-dimensional sedimentary basins is quantified, which can significantly improve the computational efficiency of the conventional Monte Carlo method. Finally, combined with a building seismic vulnerability database, pre-earthquake prediction and post-disaster assessment of building complexes in sedimentary basins are performed, demonstrating significant application value.

[0070] A seismic damage assessment system for sedimentary basin architectural complexes, used to implement the aforementioned seismic damage assessment method for sedimentary basin architectural complexes, comprising:

[0071] Data acquisition module: Determines the target site and feature input parameters, selects sedimentary basins as the target site, determines the seismic frequency bands of sedimentary basins in actual earthquakes, the site medium property parameters and their value ranges of sedimentary basins, and uses them as feature input parameters in the basic dataset of artificial neural networks;

[0072] Model building module: A three-dimensional sedimentary basin model is built and the seismic response is obtained. The three-dimensional sedimentary basin model is built and the feature input parameters are specified as the incident frequency of bedrock seismic waves, the ratio of shear wave velocity inside and outside the sedimentary basin, the damping ratio and Poisson's ratio. Each feature input parameter is randomly selected within a given value range. The seismic response of the three-dimensional sedimentary basin model under any set of random numbers is solved by the fast multipole boundary element method. The seismic responses under multiple sets of random numbers are composed of an elementary dataset.

[0073] Dataset Improvement Module: An improved dataset is established. Based on the elementary dataset, the relative position of the sedimentary basin surface is also used as a feature input parameter. The improved dataset uses the incident frequency of bedrock seismic waves, the material properties of the medium in the sedimentary basin, and the relative position of the sedimentary basin surface as feature input parameters, and the displacement amplification factor (DAF) at each position of the sedimentary basin surface as feature output parameters.

[0074] Optimization Module: Optimizes the artificial neural network; selects the artificial neural network structure, divides the improved dataset to obtain a training set, a test set, and a validation set, trains the artificial neural network based on the training set, introduces the differential evolution-particle swarm optimization algorithm to optimize the initial weights and thresholds during the training process, performs parameter selection and network accuracy testing on the artificial neural network based on the test set and validation set, and uses the artificial neural network with the best test performance as a data-driven surrogate model to solve the target problem.

[0075] Sample acquisition module: Acquires samples, establishes a real-world case model to solve the target problem, uses the trained neural network as a proxy model to analyze the uncertainty of seismic response in three-dimensional sedimentary basins, solves the basin seismic response corresponding to each sample in the target problem, and quickly provides the required samples for Monte Carlo method to analyze uncertainty problems;

[0076] Parameter Determination Module: Seismic motion parameter determination. Based on the actual sedimentary basin conditions, it sets uncertain parameters of the site medium, their probability distributions, and value ranges. It determines whether there are samples in the basic dataset that overlap with the uncertain features. If so, it extracts the results from the corresponding improved dataset. If not, it calls the surrogate model to calculate the uncertain feature combinations sequentially, obtaining the statistical moments of the sedimentary basin surface displacement amplification factor, velocity, and acceleration amplitude. It uses the inverse fast Fourier transform to obtain the peak acceleration statistical moments at any location on the sedimentary basin surface and the peak acceleration corresponding to any confidence level. It uses multi-dimensional results to conduct a quantitative assessment of the impact of site medium parameter uncertainties on three-dimensional sedimentary basin seismic motion.

[0077] Execution module: Prediction and assessment. Based on actual needs, select peak ground acceleration of sedimentary basin surface ground corresponding to different confidence levels, and combine local building seismic vulnerability curve database or vulnerability analysis to carry out seismic damage prediction and assessment of sedimentary basin building clusters considering the uncertainty of site medium parameters.

[0078] An information data processing terminal is used to implement the above-mentioned method for assessing seismic damage to building complexes in sedimentary basins.

[0079] A computer-readable storage medium includes instructions that, when executed on a computer, cause the computer to perform the aforementioned method for assessing seismic damage to sedimentary basin ensembles.

[0080] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented, in whole or in part, as a computer program product, the computer program product includes one or more computer instructions. When the computer program instructions are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)).

[0081] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Any simple modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention shall fall within the scope of the technical solution of the present invention.

Claims

1. A method for evaluating seismic damage of a sedimentary basin building group, characterized by, The application relates to a method for predicting and evaluating seismic hazard of a three-dimensional sedimentary basin, and belongs to the field of earthquake engineering. S1, determining target sites and characteristic input parameters: A sedimentary basin is selected as a target site, the seismic vibration frequency band of the sedimentary basin in actual earthquakes, sedimentary basin site medium attribute parameters and value ranges are determined, and the parameters are used as characteristic input parameters in a basic data set of an artificial neural network; S2, establishing a three-dimensional sedimentary basin model and obtaining a seismic response: A three-dimensional sedimentary basin model is established, the characteristic input parameters are specified, random numbers are taken in a given value range of each characteristic input parameter, a fast multi-pole boundary element method is used to solve the seismic response of the three-dimensional sedimentary basin model under any group of random numbers, and the seismic responses under multiple groups of random numbers form a primary data set; S3, establishing an improved data set: On the basis of the primary data set, a relative position of a sedimentary basin surface is also used as a characteristic input parameter, an improved data set is established by taking bedrock seismic wave incidence frequency, sedimentary basin internal medium material properties and sedimentary basin surface relative position as characteristic input parameters and taking a displacement amplification factor DAF of each position of a sedimentary basin surface as a characteristic output parameter; S4, optimizing an artificial neural network; S5, obtaining samples: A real case model for solving a target problem is established, the trained neural network is used as an analysis agent model for three-dimensional sedimentary basin seismic response uncertainty problems, the corresponding basin seismic response of each sample in the target problem is solved, and the required samples are quickly provided for the Monte Carlo method for analyzing uncertainty problems; S6, determining seismic vibration parameters: According to actual sedimentary basin conditions, site medium uncertainty parameters, probability distribution and value ranges are set, it is judged whether there is a sample coinciding with the uncertainty characteristics in the basic data set, if there is, the result is extracted from the corresponding improved data set, if not, the agent model is called to calculate the uncertainty characteristic combination in sequence, statistical moments of sedimentary basin surface displacement amplification factors, velocities and acceleration amplitudes are obtained, inverse fast Fourier transform is used to obtain statistical moments of peak acceleration of the sedimentary basin surface at any position and peak acceleration corresponding to any confidence degree, and multi-dimensional results are used to carry out quantitative evaluation on the influence of site medium parameter uncertainty on three-dimensional sedimentary basin seismic vibration; S7, prediction and evaluation: According to actual needs, sedimentary basin surface seismic peak acceleration corresponding to different confidence degrees is selected, a local building earthquake vulnerability curve database or vulnerability analysis is combined, and sedimentary basin building group seismic hazard prediction and evaluation considering site medium parameter uncertainty are carried out.

2. The method of claim 1, wherein, In S2, the characteristic input parameters are specified as bedrock seismic wave incidence frequency, sedimentary basin internal shear wave velocity ratio, damping ratio and Poisson's ratio.

3. The method according to claim 2, wherein In S4, an artificial neural network structure is selected, the improved data set is divided to obtain a training set, a test set and a verification set, artificial neural network training is carried out based on the training set, a differential evolution-particle swarm algorithm is introduced in the training process to optimize initial weights and thresholds, parameters of the artificial neural network are selected and network accuracy is tested based on the test set and the verification set, and the artificial neural network with the best test performance is used as a data-driven agent model for solving a target problem.

4. A system for assessing seismic damage to a sedimentary basin architecture, the system comprising: a seismic data acquisition system; a seismic data processing system; a seismic data analysis system; and a seismic data visualization system. The application further discloses a method for predicting and evaluating seismic hazard of a three-dimensional sedimentary basin. The data acquisition module: determining the target site and characteristic input parameters, selecting a sedimentary basin as the target site, determining the ground motion frequency band of the sedimentary basin in actual earthquakes, the sedimentary basin site medium attribute parameters and the value range, and taking them as the characteristic input parameters in the artificial neural network basic data set; The model establishment module: establishing a three-dimensional sedimentary basin model and obtaining the seismic response, establishing a three-dimensional sedimentary basin model, and specifying the characteristic input parameters, taking random numbers in the given value range for each characteristic input parameter, and solving the seismic response of the three-dimensional sedimentary basin model under any group of random numbers by using the fast multi-pole boundary element method, and the seismic responses under multiple groups of random numbers form the elementary data set; The data set improvement module: establishing an improved data set, on the basis of the elementary data set, taking the relative position of the sedimentary basin surface as a characteristic input parameter, establishing an improved data set taking the bedrock seismic wave incident frequency, the sedimentary basin internal medium material attribute, and the sedimentary basin surface relative position as the characteristic input parameters, and taking the displacement amplification factor DAF of each position on the sedimentary basin surface as the characteristic output parameter; The optimization module: optimizing the artificial neural network; The sample acquisition module: acquiring samples, establishing a real case model to solve the target problem, taking the trained neural network as a proxy model for analyzing the uncertainty problem of the three-dimensional sedimentary basin seismic response, solving the basin seismic response corresponding to each sample in the target problem, and quickly providing the required samples for the Monte Carlo method for analyzing the uncertainty problem; The parameter determination module: ground motion parameter determination, setting the site medium uncertainty parameters and their probability distribution and value range according to the actual sedimentary basin conditions, judging whether there are samples in the basic data set that coincide with the uncertain characteristics, if there are, extracting the results from the corresponding improved data set, if not, calling the proxy model to calculate the uncertain characteristic combinations one by one, obtaining the statistical moments of the sedimentary basin surface displacement amplification factor, velocity, and acceleration amplitude; using the inverse fast Fourier transform to obtain the peak acceleration statistical moment of the sedimentary basin surface at any position and the peak acceleration corresponding to any confidence, and using the multi-dimensional results to carry out quantitative evaluation of the influence of site medium parameter uncertainty on three-dimensional sedimentary basin ground motion; The execution module: prediction and evaluation, selecting the sedimentary basin surface ground motion peak acceleration corresponding to different confidence according to actual needs, combining with the local building seismic vulnerability curve database or vulnerability analysis, and carrying out sedimentary basin building group seismic damage prediction and evaluation considering the uncertainty of site medium parameters.

5. The system for evaluating the damage of a sedimentary basin building complex according to claim 4, wherein In the model establishment module, the characteristic input parameters are specified as the bedrock seismic wave incident frequency, the internal and external shear wave velocity ratio of the sedimentary basin, the damping ratio, and the Poisson's ratio.

6. The system for evaluating the damage of a sedimentary basin building complex according to claim 5, wherein The optimization process of the optimization module is specifically: selecting an artificial neural network structure, dividing the improved data set to obtain a training set, a test set and a validation set, training the artificial neural network based on the training set, introducing a differential evolution-particle swarm algorithm to optimize the initial weight and threshold value during the training process, selecting parameters and testing the accuracy of the artificial neural network based on the test set and the validation set, and taking the artificial neural network with the best test performance as a data-driven agent model for solving the target problem.

7. An information data processing terminal, characterized by The method for evaluating the earthquake damage of the sedimentary basin building group according to any one of claims 1-3.

8. A computer-readable storage medium, characterized in that, The computer program product comprises instructions which, when executed on a computer, cause the computer to perform the method for evaluating the earthquake damage of the sedimentary basin building group according to any one of claims 1-3.

Images