Ground vibration simulation method, system, equipment and medium

Abstract

The invention discloses a ground vibration simulation method, system and device and a medium, and relates to the field of seismic analysis, and the method comprises the steps: obtaining a plurality of seismic oscillation records; performing data filling on each seismic oscillation record by adopting a random finite fault method to obtain a plurality of sample seismic oscillation records; determining a plurality of condition parameters of each sample seismic oscillation record to obtain a multi-label vector of each sample seismic oscillation record; constructing a seismic oscillation simulation model based on the conditional denoising diffusion probability model, each sample seismic oscillation record and the multi-label vector of each sample seismic oscillation record; and carrying out seismic oscillation simulation under different seismic scenes based on the seismic oscillation simulation model. According to the invention, the complexity of the ground motion simulation model is reduced, and the precision of ground motion simulation is improved at the same time.

Description

Technical Field

[0001] The present application relates to the field of earthquake analysis, and in particular to a method, system, device and medium for simulating earthquake motion based on a multi-label conditional denoising diffusion probability model. Background Art

[0002] In seismic analysis of critical projects, incorporating appropriate ground motion records into dynamic seismic response analysis to assess the seismic demands of structures is crucial. However, relying solely on recorded ground motions is often insufficient for more accurate or detailed structural analysis. Therefore, supplementing this with simulated earthquake motions is crucial.

[0003] Currently, most methods for generating artificial seismic motions are based on physical models. These models reflect the earthquake source and propagation mechanism and usually require precise analytical equations and physical parameters. In addition, there are methods based on target spectra and random parameters, such as the trigonometric series method. These methods require certain characteristic parameters, including evolution intensity, characteristic frequency, and effective duration, to describe the earthquake time course. Some modified models attempt to introduce more parameters to reflect more complex physical mechanisms, such as fault rupture and slip mechanisms, seismic wave propagation dynamics, and site reflection effects. This helps to more comprehensively describe the factors affecting seismic motion.

[0004] However, these additional parameters often increase the difficulty of solving the modified equations and the cost of data fitting, sometimes even leading to problems such as unsolvable high-dimensional equations and data overfitting. At the same time, considering randomness also makes the new model more complex.

[0005] Deep learning, a cutting-edge tool in artificial intelligence, has been widely applied in the field of seismic simulation. However, due to factors such as the scarcity of earthquake activity, limited recording locations, and specific site conditions, building a good deep learning database presents numerous challenges. These challenges can lead to potential biases in the database, compromising the effectiveness of artificial intelligence algorithms. Summary of the Invention

[0006] The purpose of this application is to provide a seismic motion simulation method, system, device and medium, which can reduce the complexity of the seismic motion simulation model while improving the accuracy of seismic motion simulation.

[0007] To achieve the above objectives, this application provides the following solutions:

[0008] In a first aspect, the present application provides a method for simulating earthquake motion, comprising:

[0009] Obtain multiple ground motion records;

[0010] The random finite fault method is used to fill in the data of each earthquake motion record to obtain multiple sample earthquake motion records;

[0011] Determine multiple conditional parameters of each sample seismic record to obtain a multi-label vector for each sample seismic record;

[0012] A seismic motion simulation model is constructed based on the conditional denoising diffusion probability model, each sample seismic motion record, and the multi-label vector of each sample seismic motion record.

[0013] Based on the earthquake motion simulation model, earthquake motion simulations under different earthquake scenarios are performed.

[0014] Optionally, obtaining multiple seismic motion records specifically includes: collecting existing seismic motion related data, screening seismic motion related data whose earthquake moment magnitude is within a set range and whose closest distance from the earthquake source to the rupture zone is greater than a set threshold, to obtain multiple seismic motion records.

[0015] Optionally, the plurality of sample seismic records include a plurality of seismic records and seismic records after data filling.

[0016] Optionally, multiple conditional parameters of each sample seismic record are determined to obtain a multi-label vector for each sample seismic record, specifically including:

[0017] For any sample seismic record, determining a plurality of conditional parameters of the sample seismic record;

[0018] Each condition parameter of the sample earthquake record is subjected to one-hot encoding and position encoding to obtain a multi-label vector of the sample earthquake record.

[0019] Optionally, the multiple conditional parameters are average shear wave velocity, fault type, earthquake moment magnitude, and the closest distance from the earthquake source to the rupture zone.

[0020] Optionally, the fault type of the sample ground motion record is a normal fault, a reverse fault or a strike-slip fault;

[0021] The seismic moment magnitude is divided into labels at intervals of 0.5 to determine the seismic moment magnitude of the sample ground motion record;

[0022] The closest distance from the earthquake source to the rupture zone is divided into sections to determine the closest distance from the earthquake source to the rupture zone of the sample earthquake motion record.

[0023] Optionally, a seismic motion simulation model is constructed based on the conditional denoising diffusion probability model, each sample seismic motion record, and a multi-label vector of each sample seismic motion record, specifically including:

[0024] A hybrid database is constructed based on each sample seismic record and the multi-label vector of each sample seismic record;

[0025] Based on the multi-label vector and the conditional denoising diffusion probability model, a multi-label conditional denoising diffusion probability model is constructed;

[0026] The multi-label conditional denoising diffusion probability model is used to perform deep learning on the hybrid database to obtain a seismic simulation model.

[0027] In a second aspect, the present application provides a seismic simulation system, comprising:

[0028] A data acquisition module, used to acquire multiple earthquake motion records;

[0029] A data filling module is used to fill the data of each earthquake motion record using a random finite fault method to obtain multiple sample earthquake motion records;

[0030] a label marking module, used to determine multiple conditional parameters of each sample seismic record to obtain a multi-label vector for each sample seismic record;

[0031] A model building module is used to build a seismic motion simulation model based on a conditional denoising diffusion probability model, each sample seismic motion record, and a multi-label vector of each sample seismic motion record;

[0032] The seismic motion simulation module is used to simulate seismic motion under different earthquake scenarios based on the seismic motion simulation model.

[0033] In a third aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-mentioned earthquake motion simulation method.

[0034] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which implements the above-mentioned earthquake motion simulation method when executed by a processor.

[0035] According to the specific embodiments provided in this application, this application has the following technical effects:

[0036] The present application provides a seismic motion simulation method, system, equipment and medium, which effectively alleviates the problem of scarce seismic motion records through the random finite fault method, and constructs a seismic motion simulation model based on a conditional denoising diffusion probability model and a multi-label vector, and integrates multiple conditional parameters into the deep learning model, thereby realizing the recognition and reproduction of multi-label features of seismic motion under different earthquake scenarios, reducing the complexity of the seismic motion simulation model, and improving the accuracy of seismic motion simulation. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0038] Figure 1 This is a diagram of an application environment of a seismic simulation method in one embodiment of the present application;

[0039] Figure 2 A schematic flow chart of a method for simulating earthquake motion provided in one embodiment of the present application;

[0040] Figure 3 A schematic diagram of the hierarchical relationship between conditional parameters in an embodiment of the present application;

[0041] Figure 4 Schematic diagram of the encoding process in one embodiment of the present application;

[0042] Figure 5 A schematic diagram of the basic architecture of a multi-label conditional denoising diffusion probability model provided in one embodiment of the present application;

[0043] Figure 6 A schematic diagram of the U-Net network architecture in the multi-label conditional denoising diffusion probability model provided in one embodiment of the present application;

[0044] Figure 7 Schematic diagram of the forward diffusion process of the multi-label conditional denoising diffusion probability model provided in one embodiment of the present application;

[0045] Figure 8 Schematic diagram of the inverse denoising process of the multi-label conditional denoising diffusion probability model provided in one embodiment of the present application;

[0046] Figure 9 A schematic diagram of the functional modules of a seismic simulation system provided in one embodiment of the present application. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0048] Although studies have demonstrated the potential of some deep learning models, such as Generative Adversarial Networks (GAN) and Variational Auto-Encoders (VAE), in ground motion simulation, their performance in specific earthquake frequency bands still needs to be improved. Therefore, stronger neural networks that capture and reproduce the time and frequency characteristics of ground motion are needed. Diffusion models have shown excellent performance in various practical applications. Therefore, this application introduces a diffusion model. The diffusion model uses different initial states and each transition of a simple distribution to simulate different samples, which means that it is challenging to control the content of the output image due to the randomness of the initial state and transition. To solve this problem, Lu et al. (2022) introduced the concept of conditional denoising diffusion probabilistic model (cDenoising Diffusion Probabilistic Model, cDDPM), which allows content control through a single conditional label Y. However, for seismic motion simulation, using a single label Y is not sufficient to satisfy the average shear wave velocity V based on the top 30 meters. S30 , fault type F, earthquake moment magnitude M W and the shortest distance R from the earthquake source to the rupture zone rup Therefore, this application further proposes a multi-label conditional embedding method to transform cDDPM to achieve ground motion simulation under more complex and diverse earthquake scenarios.

[0049] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0050] The earthquake simulation method provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be set up separately, integrated on the server 104, or placed on the cloud or other servers. The terminal 102 can send seismic motion records to the server 104. After receiving the seismic motion records, the server 104 constructs a seismic motion simulation model and performs seismic motion simulation under different earthquake scenarios. The server 104 can feedback the seismic motion simulation results under different earthquake scenarios to the terminal 102. In addition, in some embodiments, the seismic motion simulation method can also be implemented separately by the server 104 or the terminal 102.

[0051] Terminal 102 may include, but is not limited to, various desktop computers, laptops, smartphones, tablet computers, IoT devices, and portable wearable devices. IoT devices may include smart speakers, smart TVs, smart air conditioners, and smart car devices. Portable wearable devices may include smart watches, smart bracelets, and head-mounted devices. Server 104 may be implemented as a standalone server or a server cluster consisting of multiple servers, or may be a cloud server.

[0052] In an exemplary embodiment, Figure 2 As shown, a method for simulating earthquake motion is provided. The method is executed by a computer device. Specifically, the method can be executed by a computer device such as a terminal or a server alone, or by a terminal and a server together. In the embodiment of the present application, the method is applied to Figure 1 The server 104 in the example is used for explanation, including the following steps 201 to 205.

[0053] Step 201: Acquire multiple earthquake motion records.

[0054] Specifically, existing seismic motion related data are collected, and seismic motion related data whose earthquake moment magnitude is within a set range and whose closest distance from the earthquake source to the rupture zone is greater than a set threshold are screened to obtain multiple seismic motion records.

[0055] When selecting earthquake motion records, the earthquake moment magnitude M W The range is 5.0~8.0, and the closest distance R from the earthquake source to the rupture zone is rup Less than 100 km, while excluding earthquake events with less than 3 records within 100 km and aftershock records to ensure the quality and validity of the seismic motion records.

[0056] In step 202, a random finite fault method is used to fill in data on each seismic motion record to obtain a plurality of sample seismic motion records. The plurality of sample seismic motion records includes the seismic motion records and the seismic motion records after data filling.

[0057] By combining traditional seismology with artificial intelligence algorithms, we can fully leverage the advantages of earthquake physics and data characteristics to extract more information from ground motion signal characteristics. Therefore, this application uses a random finite fault method to simulate ground motion, making up for the less distributed areas of ground motion records, effectively filling data gaps, and enhancing the comprehensiveness and representativeness of the dataset.

[0058] The random finite fault method decomposes a large earthquake (i.e., a ground motion record) into multiple sub-faults, each of which is equivalent to a small point source. The ground motion time history of each sub-fault is calculated using a random point source model. The time difference is then taken into account and superimposed to obtain the ground motion time history for the entire fault.

[0059] The random point source model is calculated according to the following formula:

[0060] E(M0,R,f)=S(M0,f)·P(R,f)·G(f)·I(f);

[0061] Where M0 is the seismic moment, R is the focal distance, f is the frequency, S(M0,f) is the Fourier spectrum of seismic motion, which is defined as the source spectrum, P(R,f) is the path transfer function, G(f) is the site transfer function, and I(f) is the seismic motion type conversion function.

[0062] Step 203 : determining multiple conditional parameters of each sample earthquake record to obtain a multi-label vector for each sample earthquake record.

[0063] In an exemplary embodiment, for any sample seismic record, a plurality of condition parameters of the sample seismic record are determined.

[0064] Among them, multiple conditional parameters are the average shear wave velocity (here the average shear wave velocity V in the top 30 meters) S30 ), fault type F, earthquake moment magnitude M W and the closest distance R from the earthquake source to the rupture zone rup By marking key condition information in sample seismic records, we ensure that the seismic simulation model can accurately understand and simulate different earthquake scenarios.

[0065] The fault type of the sample seismic record is a normal fault, a reverse fault, or a strike-slip fault. The seismic moment magnitude is labeled at intervals of 0.5 to determine the seismic moment magnitude of the sample seismic record. The closest distance from the earthquake source to the rupture zone is segmented to determine the closest distance from the earthquake source to the rupture zone of the sample seismic record.

[0066] like Figure 3 As shown in the figure, the fault type F is marked according to the three different fault types: normal fault, reverse fault, and strike-slip fault; the earthquake moment magnitude M W The labels are divided at intervals of 0.5; considering the closest distance R from the earthquake source to the rupture zone rup The difference in the value of earthquake motion records in different distance ranges is marked in sections. rup ≤10km is marked at 5km intervals, 10km <R rup ≤40km is marked at intervals of 10km, 40km <R rup ≤100km is marked at intervals of 20km; and the average shear wave velocity V in the top 30 meters S30The local site conditions are classified according to the National Earthquake Hazards Reduction Program (NEHRP) standards, NEHRP B / C (rock / compact soil) and NEHRP D (hard soil).

[0067] Each conditional parameter of the sample seismic record is one-hot encoded and positionally encoded to obtain a multi-label vector for the sample seismic record. This technique is used to construct a multi-label conditional embedding model, which is then integrated into a subsequent deep learning model, enabling the seismic simulation model to flexibly adjust simulation results based on specific conditions.

[0068] like Figure 4 As shown in the figure, during the encoding process, each conditional parameter is first encoded using one-hot encoding, that is, for each category, an embedding vector with a length equal to the total number of categories is created. In this vector, only the position corresponding to the category is 1, and the rest of the positions are 0. The length of the vector is equal to the number of conditional parameters. For example, y1 is used to represent the average shear wave velocity V in the top 30 meters. S30 The unique heat encoding of y2 represents the unique heat encoding of fault type F, and y3 represents the earthquake moment magnitude M W The unique heat encoding, y4 represents the closest distance R from the earthquake source to the rupture zone rup One-hot encoding of .

[0069] In order to further reflect the hierarchical relationship between conditional information, the following formula is used to positionally encode conditional parameters using sine and cosine functions of different frequencies:

[0070]

[0071] Among them, pos is the position, i is the dimension, d model is the dimension of the embedding vector, which represents the spatial location of the four conditional parameters.

[0072] The process of position encoding is as follows:

[0073] (1) Initialize the position encoding matrix. Create a zero matrix of size [sequence length, embedding vector dimension] to store the position encoding of each position.

[0074] (2) Calculate the numerator and denominator of the positional encoding. Numerator: For each position pos, create a one-dimensional array with the same dimension as the embedding vector, and each element in the array is pos. Denominator: For each dimension i of the embedding vector, calculate 10000^(2i / d model ) and creates a one-dimensional array of the same length as the sequence.

[0075] (3) Applying sine and cosine functions: According to the calculation formulas for even and odd dimensions, sine and cosine functions are applied to the numerator and denominator to obtain the position code of each position in each dimension.

[0076] (4) Combined position coding: Combine the calculated position coding in order to form a matrix of size [sequence length, embedding vector dimension]. pos1 is used to represent the average shear wave velocity V in the top 30 meters. S30 pos2 represents the position code of fault type F, pos3 represents the earthquake moment magnitude M W Position code, pos4 represents the shortest distance R from the earthquake source to the rupture zone rup Position encoding.

[0077] Then, the unique hot encoding and position encoding are fused to obtain a multi-label vector for each sample earthquake record. In the multi-label vector, c1 is used to represent the average shear wave velocity V at the top 30 meters. S30 The label of the fault type F is c2, and the label of the earthquake moment magnitude M is c3. W The label, c4 represents the closest distance R from the earthquake source to the rupture zone rup .

[0078] Step 204 : constructing a seismic motion simulation model based on the conditional denoising diffusion probability model, each sample seismic motion record, and the multi-label vector of each sample seismic motion record.

[0079] In an exemplary embodiment, step 204 includes the following steps (1) to (3):

[0080] (1) A hybrid database is constructed based on each sample seismic motion record and its multi-label vector. By integrating seismic motion records and records compensated by the random finite fault method, a hybrid database suitable for deep learning algorithms is constructed, providing rich and diverse learning materials for seismic motion simulation model training.

[0081] (2) Based on the multi-label vector and the conditional denoising diffusion probability model, a multi-label conditional denoising diffusion probability model is constructed.

[0082] (3) Using the multi-label conditional denoising diffusion probability model to perform deep learning on the hybrid database, a seismic simulation model is obtained, thereby improving the learning efficiency and accuracy of the seismic simulation model.

[0083] The basic architecture of the multi-label conditional denoising diffusion probability model is as follows Figure 5 As shown, its core usage is as follows Figure 6The U-Net network architecture shown in the figure. The sample seismic records in this application are input into the multi-label conditional denoising diffusion probability model in image form. During the denoising process, the multi-label conditional denoising diffusion probability model continuously adds noise to the image, resulting in a completely noisy image. The first U-Net records the entire process of adding noise. During the denoising process, the second U-Net predicts the noise added to each completely noisy image with four conditional parameters, removes the noise, and restores the original image.

[0084] like Figure 7 As shown in Figure 2, the forward diffusion process of the multi-label conditional denoising diffusion probability model during training is described by the following formula:

[0085]

[0086] like Figure 8 As shown in Figure 2, the inverse process of the multi-label conditional denoising diffusion probability model during training is described by the following formula:

[0087]

[0088] The loss function of the multi-label conditional denoising diffusion probability model during training is described by the following formula:

[0089]

[0090] Among them, x0~q(x0) is the given ground motion data distribution, I is the unit matrix, {β} t＝1:T represents a constant controlling the noise step size or learning variance schedule, p θ (x 0:T ) is the reverse denoising process, which is a process from x T Markov chain sampling backward to x0: x t represents the diffusion process of each iteration of the diffusion model, t is the number of iterations, y represents the multi-label vector, which consists of one-hot encoding and position encoding, p θ It is defined as following a Gaussian distribution, m θ represents the mean, ∑ θ represents the variance, is the loss function value, ∈ θ is the noise injected during the noise addition process.

[0091] In order to ensure the learning quality of the multi-label conditional denoising diffusion probability model, this application compares the learning results of the multi-label conditional denoising diffusion probability model with previous earthquake records, mixed databases, and seismic motion prediction equation results, and then verifies the deep learning results to ensure that the learning simulation results of the multi-label conditional denoising diffusion probability model are consistent with both visual logic and physical laws.

[0092] Specifically, seismic records from at least three different earthquakes were selected and compared with the deep learning results in terms of both time domain and response spectrum. Secondly, the deep learning results were compared with the hybrid database in terms of average power spectrum and response spectrum. Finally, at least three existing seismic prediction equations were selected and compared with the response spectrum of the deep learning results. The average power spectrum was calculated according to the following formula:

[0093]

[0094] Among them, S xx (t,f) is the time-frequency power spectrum, S xx (f) is the average time-frequency power spectrum.

[0095] Step 205: Perform different earthquake scenarios (V S30 , F, M W and R rup ) under the earthquake motion simulation.

[0096] This application effectively alleviates the problem of scarce seismic motion records through the random finite fault method, and at the same time proposes a multi-label conditional embedding method to improve cDDPM, integrating conditional parameters into the deep learning model, and realizing the recognition and reproduction of multi-label features of seismic motion under different earthquake scenarios, providing strong technical support for earthquake science research and disaster prevention and mitigation.

[0097] Based on the same inventive concept, embodiments of the present application also provide a seismic motion simulation system for implementing the aforementioned seismic motion simulation method. The solution provided by this system is similar to the solution described in the aforementioned method. Therefore, the specific limitations in one or more of the seismic motion simulation system embodiments provided below can be found in the above-described limitations on the seismic motion simulation method and will not be further elaborated here.

[0098] In an exemplary embodiment, Figure 9 As shown, a seismic motion simulation system is provided, including: a data acquisition module 901 , a data filling module 902 , a label marking module 903 , a model building module 904 and a seismic motion simulation module 905 .

[0099] The data acquisition module 901 is used to acquire multiple earthquake records.

[0100] The data filling module 902 is used to fill the data of each earthquake motion record using a random finite fault method to obtain multiple sample earthquake motion records.

[0101] The label marking module 903 is used to determine multiple conditional parameters of each sample seismic record to obtain a multi-label vector for each sample seismic record.

[0102] The model building module 904 is used to build a seismic motion simulation model based on the conditional denoising diffusion probability model, each sample seismic motion record and the multi-label vector of each sample seismic motion record.

[0103] The earthquake motion simulation module 905 is used to perform earthquake motion simulation under different earthquake scenarios based on the earthquake motion simulation model.

[0104] In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0105] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0106] In an exemplary embodiment, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0107] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0108] In this application, all actions to obtain signals, information or data are carried out in compliance with the relevant data protection laws and policies of the country where they are located and with the authorization given by the owner of the corresponding device.

[0109] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0110] The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processors involved in the various embodiments provided herein may include, but are not limited to, general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic units, data processing logic units based on quantum computing, and the like.

[0111] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0112] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above examples is only intended to help understand the method and core concept of this application. At the same time, for those skilled in the art, based on the concept of this application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for simulating earthquake motion, characterized in that: The earthquake simulation method comprises: Get multiple ground motion records; The random finite fault method is used to fill the data of each earthquake motion record to obtain multiple sample earthquake motion records; Determine multiple condition parameters of each sample ground motion record to obtain a multi-label vector of each sample ground motion record; Based on the conditional denoising diffusion probability model, each sample ground motion record and the multi-label vector of each sample ground motion record, a ground motion simulation model is constructed; Based on the earthquake motion simulation model, earthquake motion simulations are performed under different earthquake scenarios.

2. The earthquake simulation method according to claim 1, characterized in that: Get multiple earthquake records, including: Collect existing seismic motion related data, screen seismic motion related data whose earthquake moment magnitude is within a set range and whose closest distance from the epicenter to the rupture zone is greater than a set threshold, to obtain multiple seismic motion records.

3. The earthquake simulation method according to claim 1, characterized in that: The plurality of sample seismic motion records include a plurality of seismic motion records and seismic motion records after data filling.

4. The earthquake simulation method according to claim 1, characterized in that: Determine multiple conditional parameters of each sample seismic record to obtain a multi-label vector for each sample seismic record, including: For any sample seismic record, determining a plurality of conditional parameters of the sample seismic record; Each condition parameter of the sample seismic record is uniquely encoded and positionally encoded to obtain a multi-label vector of the sample seismic record.

5. The earthquake simulation method according to claim 4, characterized in that: The multiple conditional parameters are the average shear wave velocity, fault type, earthquake moment magnitude, and the closest distance from the epicenter to the rupture zone.

6. The earthquake simulation method according to claim 5, characterized in that: The fault type of the sample ground motion record is a normal fault, a reverse fault or a strike-slip fault; The seismic moment magnitude is divided into labels at intervals of 0.5 to determine the seismic moment magnitude of the sample ground motion record; The shortest distance from the earthquake source to the rupture zone is divided into sections to determine the shortest distance from the earthquake source to the rupture zone of the sample earthquake motion record.

7. The earthquake simulation method according to claim 1, characterized in that: Based on the conditional denoising diffusion probability model, each sample ground motion record and the multi-label vector of each sample ground motion record, a ground motion simulation model is constructed, which specifically includes: A hybrid database is constructed according to each sample seismic record and a multi-label vector of each sample seismic record; Based on the multi-label vector and the conditional denoising diffusion probability model, a multi-label conditional denoising diffusion probability model is constructed; The multi-label conditional denoising diffusion probability model is used to perform deep learning on the hybrid database to obtain a seismic simulation model.

8. A seismic simulation system, applied to the seismic simulation method according to any one of claims 1 to 7, characterized in that: The earthquake simulation system comprises: A data acquisition module, used to acquire multiple earthquake records; A data filling module is used to fill the data of each earthquake motion record by using a random finite fault method to obtain multiple sample earthquake motion records; A label marking module, used for determining a plurality of conditional parameters of each sample seismic record to obtain a multi-label vector of each sample seismic record; A model building module, used to build a seismic motion simulation model based on a conditional denoising diffusion probability model, each sample seismic motion record and a multi-label vector of each sample seismic motion record; The seismic motion simulation module is used to perform seismic motion simulation under different earthquake scenarios based on the seismic motion simulation model.

9. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the earthquake motion simulation method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the earthquake motion simulation method according to any one of claims 1 to 7 is implemented.

Images