Earthquake motion intelligent synthesis method considering multi-classification features and related equipment

Abstract

The embodiment of the invention provides an intelligent seismic motion synthesis method considering multi-classification features and related equipment, and seismic motion synthesis data with different feature attributes can be generated. The method comprises the steps that a target three-direction seismic wave data set is acquired, each piece of seismic wave data in the target three-direction seismic wave data set comprises N data points, and N is an integer larger than or equal to 1; constructing a target seismic data matrix based on the target three-direction seismic wave data set; performing principal component analysis on the target seismic data matrix to obtain target one-dimensional seismic wave data corresponding to each piece of seismic wave data in the target three-direction seismic wave data; and splicing the target one-dimensional seismic wave data and the random noise corresponding to the target one-dimensional seismic wave data, and inputting the spliced data into a pre-trained target generator combination to obtain seismic motion synthesis data.

Description

Technical Field

[0001] The present invention relates to the field of earthquake motion, and in particular to an intelligent synthesis method of earthquake motion considering multiple classification features and related equipment. Background Art

[0002] When conducting engineering earthquake simulation, it is often necessary to input the characteristic seismic waveform of the site in advance. Due to the lack of offshore earthquake records, reasonable seismic motion generation is of great significance for the analysis of engineering earthquake dynamic response. Real seismic records often show obvious time-frequency non-stationarity. Synthetic seismic motion waveforms should consider many constraints, such as frequency, amplitude, duration, response spectrum, etc. In addition, different site conditions, seismic motion parameters and other classification characteristics also have an important impact on seismic motion. For example, near-fault pulse waves and far-field fluctuations have obvious differences. Therefore, the seismic motion synthesis method should consider the time-domain non-stationarity and classification characteristics of seismic waves.

[0003] The methods proposed so far mainly include two categories: random finite fault method and engineering synthesis method. The parameter determination of random finite fault method is more complicated and the reproducibility is poor; the engineering synthesis method is highly dependent on statistical mathematical models and requires statistical analysis of a large number of seismic motion records in advance, but statistical laws often cannot be fully expressed by mathematical models. Summary of the invention

[0004] The present invention provides a method and device for intelligent synthesis of seismic motions taking into account multi-classification characteristics, which can generate synthetic seismic motion data with different characteristic attributes, and provide a scientific basis for earthquake dynamic response simulation and earthquake load input of marine engineering.

[0005] A first aspect of the present invention provides a method for intelligent synthesis of earthquake motions taking into account multi-classification features, the method comprising:

[0006] Acquire a target three-directional seismic wave data set, wherein each seismic wave data in the target three-dimensional seismic wave data set includes N data points, where N is an integer greater than or equal to 1;

[0007] Constructing a target seismic data matrix based on the target three-directional seismic wave data set;

[0008] Performing principal component analysis on the target seismic data matrix to obtain target one-dimensional seismic wave data corresponding to each seismic wave data in the target three-directional seismic wave data;

[0009] The target one-dimensional seismic wave data and the random noise corresponding to the target one-dimensional seismic wave data are spliced ​​and input into a pre-trained target generator combination to obtain synthetic seismic motion data.

[0010] A second aspect of the present invention provides an intelligent synthesis device for earthquake motions taking into account multiple classification features, comprising:

[0011] An acquisition module, used for acquiring a target three-directional seismic wave data set, wherein each seismic wave data in the target three-dimensional seismic wave data set includes N data points, where N is an integer greater than or equal to 1;

[0012] A construction module, used for constructing a target seismic data matrix based on the target three-directional seismic wave data set;

[0013] An analysis module, used for performing principal component analysis on the target seismic data matrix to obtain target one-dimensional seismic wave data corresponding to each seismic wave data in the target three-directional seismic wave data;

[0014] The generation module is used to splice the target one-dimensional seismic wave data and the random noise corresponding to the target one-dimensional seismic wave data and input them into a pre-trained target generator combination to obtain seismic synthetic data.

[0015] A third aspect of an embodiment of the present invention provides an electronic device, including a memory and a processor, wherein the processor is used to implement the steps of the intelligent synthesis method of seismic motion considering multiple classification features as described in the first aspect when executing a computer management program stored in the memory.

[0016] A fourth aspect of an embodiment of the present invention provides a computer-readable storage medium on which a computer management program is stored. When the computer management program is executed by a processor, the steps of the intelligent synthesis method of seismic motion considering multiple classification features as described in the first aspect above are implemented.

[0017] To sum up, it can be seen that in the embodiments provided in the present application, the seismic motion data, the corresponding classification features and the target generator trained in advance by random noise input are combined to generate corresponding seismic motion synthetic data, fully learn the time domain non-stationarity and multi-classification characteristics of marine seismic motion, and generate seismic motion synthetic data with different characteristic attributes, providing a scientific basis for the simulation of marine engineering earthquake dynamic response and seismic load input. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A schematic diagram of a flow chart of an intelligent synthesis method of earthquake motion considering multiple classification features provided in an embodiment of the present application;

[0019] Figure 2 A schematic diagram of the structure of a parameterized generator combination and a parameterized discriminator combination provided by an embodiment of the present invention;

[0020] Figure 3 A schematic diagram of the convolutional neural network structure of the generator provided in an embodiment of the present invention;

[0021] Figure 4 A schematic diagram of the convolutional neural network structure of the discriminator provided in an embodiment of the present invention;

[0022] Figure 5 A schematic diagram of the change of the loss function of the parameterized generator and the parameterized discriminator provided in the embodiment of the present invention;

[0023] Figure 6 A schematic diagram of a waveform of a synthetic seismic wave generated by a trained parameterized generator provided in an embodiment of the present invention;

[0024] Figure 7 A schematic diagram showing the comparison of the accumulated energies of real seismic waves and synthetic seismic waves provided by an embodiment of the present invention;

[0025] Figure 8 A schematic diagram showing a comparison of response spectra of a real seismic wave and a synthetic seismic wave of type C provided in an embodiment of the present invention;

[0026] Fig. 9 A virtual structural diagram of an intelligent seismic synthesis device considering multiple classification features provided by an embodiment of the present invention;

[0027] Fig.10 A schematic diagram of the hardware structure of an intelligent seismic synthesis device considering multiple classification features provided by an embodiment of the present invention;

[0028] Fig.11 A schematic diagram of an embodiment of an electronic device provided by an embodiment of the present invention;

[0029] Fig.12 A schematic diagram of an embodiment of a computer-readable storage medium provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0030] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0031] The following describes the intelligent synthesis method of seismic motion considering multiple classification features from the perspective of an intelligent synthesis device for seismic motion considering multiple classification features. The intelligent synthesis device for seismic motion considering multiple classification features can be a server or a service unit in the server, without specific limitation.

[0032] See also Figure 1 , Figure 1A schematic flow chart of a method for intelligent synthesis of earthquake motions considering multiple classification features provided by an embodiment of the present invention includes:

[0033] 101. Obtain a target three-directional seismic wave data set.

[0034] In this embodiment, the intelligent synthesis device for seismic motion taking into account multi-classification features can obtain a target three-directional seismic wave data set, wherein the target three-directional seismic wave data set includes M groups of three-directional seismic wave data, and each group of three-directional seismic wave data includes N data points, wherein M and N are both integers greater than or equal to 1, and the method for obtaining the target three-directional seismic wave data is not specifically limited here.

[0035] 102. Construct a target seismic data matrix based on the target three-directional seismic wave data set.

[0036] In this embodiment, the intelligent seismic synthesis device considering multiple classification features can construct a target seismic data matrix X based on the target three-directional seismic wave data set, wherein the target seismic data matrix is ​​represented by the following formula:

[0037]

[0038] Among them, x ij is the i-th data point in the j-th group of three-directional seismic wave data, i is any one of the N data points, and j is any one of the M groups of target three-directional seismic wave data.

[0039] 103. Perform principal component analysis on the target seismic data matrix to obtain target one-dimensional seismic wave data corresponding to each seismic wave data in the target three-directional seismic wave data set.

[0040] In this step, the intelligent synthesis device for earthquake motion considering multiple classification features first performs data standardization processing on each earthquake wave data in the first group of three-directional earthquake wave data, where the first group of three-directional earthquake wave data is any group of three-directional earthquake wave data in the M groups of three-directional earthquake wave data. Specifically, the intelligent synthesis device for earthquake motion considering multiple classification features can perform data standardization processing on each earthquake wave data by the following formula:

[0041]

[0042] x′ ij For x ij After the standardized data, var(x j ) is the mean square error, is the mean,

[0043] Afterwards, the correlation coefficient matrix corresponding to each seismic wave data in the first group of three-directional seismic wave data is calculated. Specifically, the correlation coefficient matrix can be expressed by the following formula:

[0044]

[0045] in, x′ ij is the sample data after standardization.

[0046] Finally, the target one-dimensional seismic wave data corresponding to each seismic wave data in the first group of three-directional seismic wave data is determined according to the singular value decomposition method (SVD) and the correlation coefficient matrix. Specifically,

[0047] R=UDV T

[0048] Among them, U corresponds to M M-dimensional standard orthogonal vectors, V T Corresponding to N N-dimensional standard orthogonal vectors, they correspond to RR T and R T The decomposed eigenvectors of R are used to compress the number of rows and columns respectively. D corresponds to a diagonal matrix of M rows and N columns, and the diagonal elements represent the square of the eigenvalues. Based on the eigenvectors and eigenvalues, the target one-dimensional seismic wave data can be obtained.

[0049] It should be noted that the target one-dimensional seismic wave data can be used as the input value of the parameterized generator combination, X M =X M×N V N , X M is the input data of the parameterized generator combination after dimension reduction, X M×N is the original sample data, V N V T The corresponding eigenvector in .

[0050] 104. The target one-dimensional seismic wave data and the random noise corresponding to the target one-dimensional seismic wave data are spliced ​​and input into a pre-trained target generator combination to obtain seismic motion data.

[0051] In this embodiment, the intelligent seismic motion synthesis device that considers multiple classification features can divide the seismic wave data into three categories according to the characteristic parameters. For example, according to the indicators of magnitude 4 and 6, the seismic wave data can be divided into small earthquakes, medium earthquakes and large earthquakes. Then, the classification features corresponding to the target one-dimensional seismic wave data are determined according to the characteristic parameters, and the target one-dimensional seismic wave data and the random noise corresponding to the target one-dimensional seismic wave data are spliced ​​to obtain spliced ​​data, and the spliced ​​data is input into the pre-trained target generator combination to obtain seismic motion synthesis data.

[0052] The following is a detailed description of how to train the target generator combination:

[0053] Step A1: acquiring m groups of three-directional seismic wave data according to the user's operation instructions.

[0054] In this step, the intelligent seismic motion synthesis device considering multi-classification features obtains m groups of three-directional seismic wave data according to the user's operation instructions. Each seismic wave data in the m groups of three-dimensional seismic wave data includes n data points, wherein m and n are both integers greater than or equal to 1, that is, in the existing seismic wave data, a total of m groups of three-directional seismic waveform data are manually selected for training and testing.

[0055] Step A2: construct a seismic data matrix based on m groups of three-directional seismic wave data.

[0056] Step A3: perform principal component analysis on the seismic data matrix to obtain one-dimensional seismic wave data corresponding to each seismic wave data in the m groups of three-directional seismic wave data.

[0057] It should be noted that the method of constructing the seismic data matrix in step A2 is similar to the method of constructing the target seismic data matrix in step 102, and the method of determining the one-dimensional seismic wave data in step A3 is similar to the method of constructing the target one-dimensional seismic wave data in step 103. The above has been explained in detail and will not be repeated here.

[0058] Step A4: divide the one-dimensional seismic wave data into first-category seismic wave data, second-category seismic wave data and third-category seismic wave data according to preset characteristic parameters, and construct a labeled seismic motion database according to the first-category seismic wave data, the second-category seismic wave data and the third-category seismic wave data.

[0059] In this step, the intelligent synthesis device for seismic motion considering multi-classification features is based on one-dimensional seismic wave data, and divides the seismic waves into three categories according to preset characteristic parameters, namely, first-category seismic wave data, second-category seismic wave data and third-category seismic wave data. The preset characteristic parameters, for example, divide the seismic wave data into small earthquakes, medium earthquakes and large earthquakes according to the indicators of magnitude 4 and 6, thereby constructing a labeled seismic motion database.

[0060] Step A5: train the initial parameterized generator combination and discriminator combination according to the earthquake motion database to obtain the target generator combination and the target discriminator combination.

[0061] In this step, after building a labeled earthquake database, the earthquake intelligent synthesis device considering multiple classification features can train the initial parameterized generator combination and the discriminator combination according to the earthquake database to obtain the target generator combination and the target discriminator combination. For ease of understanding, the following is a detailed description of how to train the target generator combination and the target discriminator combination:

[0062] The intelligent seismic motion synthesis device considering multi-classification features trains the initial parameterized generator combination and the initial parameterized discriminator according to the seismic motion database to obtain the target generator combination and the target discriminator combination including:

[0063] A51, inputting the one-dimensional seismic wave data into the initial parameterized generator combination to obtain the first simulated seismic wave data and the second simulated seismic wave data corresponding to the target seismic wave data;

[0064] In this step, the intelligent seismic synthesis device considering multiple classification features can input the one-dimensional seismic wave data into the initial parameterized generator combination to obtain the first simulated seismic wave data and the second simulated seismic wave data corresponding to the one-dimensional seismic wave data, see Figure 2 The initial parameterized generator combination includes 6 generators. The initial parameterized generator combination considers the bidirectional mapping of three types of seismic wave classification characteristics. Specifically, the type symbol A is defined to correspond to the first type of seismic wave data, the type symbol B is defined to correspond to the second type of seismic wave data, the type symbol C is defined to correspond to the third type of seismic wave data, and the symbol "→" represents the transformation from the seismic wave on the left side of "→" to the seismic wave on the right side of "→".

[0065] Combine the following Figure 2 The following specifically explains how to obtain two simulation characteristic data:

[0066] Taking the transformation of the first type of seismic wave to the second type of seismic wave (A→B) as an example, the real one-dimensional seismic wave data X A and random noise Z A Through the generator G AB Get simulated seismic wave data X ’ B , simulated seismic wave data X ’ B and random noise Z B Then through the generator G BA Get simulated seismic wave data X ’ A ;

[0067] Combination Figure 2Referring to the A→B process, the step of inputting the one-dimensional seismic wave data into the initial parameterized generator combination also includes A→C, B→A, B→C, C→A, C→B, a total of 6 types of transformation processes.

[0068] Combined with reference Figure 3 The parameterized generators in the parameterized generator combination all adopt CNN structure, including forward mapping CNN model and reverse mapping CNN model. The input of the forward mapping CNN model is one-dimensional seismic wave data X, and then random data of a certain length is spliced ​​on X to introduce random terms; then a fully connected layer is used to adjust the length of the input data; then a convolutional pooling layer is used to downsample the input data to extract high-dimensional features; then the deconvolution layer and the convolution layer are used alternately to interpret the high-dimensional features and restore them to the set waveform length; finally, an additional convolutional layer is used to integrate multi-channel information, and the feature map is converted into a synthetic waveform through the tanh function.

[0069] Among them, the number of hidden units of the fully connected layer Gfull is the same as the length of the seismic wave data; the upsampling ratio of the deconvolution layers Gdeconv1~Gdeconv4 is 2, and the number of output layers is [4,8,16,32]; the convolution kernel length of the convolution layers Gconv1~Gconv9 is 5, the step size is 1, and the number of output layers is [4,8,16,32,32,16,8,4,1] respectively. It should be noted that the parameters of each layer are not fixed values, but should be dynamically adjusted according to the training effect to achieve the best generation effect.

[0070] Step A52, inputting the first simulated seismic wave data, the second simulated seismic wave data and the real one-dimensional seismic wave data into the corresponding seismic wave discriminators of each type to obtain the true or false prediction value corresponding to each waveform;

[0071] In this step, the input of the parameterized discriminator in the parameterized discriminator combination is seismic wave data. After the convolution layer, pooling layer and fully connected layer, the true and false prediction values ​​of the input seismic wave data are obtained. Figure 4 To explain in detail:

[0072] The input of the parameterized discriminator is the original seismic wave sequence X; then the convolution layer and the pooling layer are used alternately to extract high-dimensional features; then, the multi-channel features of the fully connected layer are integrated; finally, the fully connected layer is used and the sigmoid function is used to generate the true and false prediction value S.

[0073] Among them, the stride of the convolutional layers Dconv1~Dconv5 is 1, the convolution kernel lengths are [3,3,3,5,5] respectively, and the number of output channels is [4,8,16,32,64] respectively; the pooling receptive field and stride of the pooling layers Dpool1 and Dpool2 are both 2, and the pooling receptive field and stride of Dpool3~Dpool5 are both 4; the number of hidden units of the fully connected layers Dfull1 and Dfull2 are 128 and 1 respectively. It should be noted that the parameters of each layer are not fixed values, but should be dynamically adjusted according to the training effect to achieve the best true and false judgment effect.

[0074] Step A53, fixing the initial parameterized generator combination, and determining the loss value of the initial parameterized discriminator combination according to the generator loss function corresponding to the initial parameterized generator combination;

[0075] In this step, based on the true and false prediction values ​​corresponding to each waveform, the Adam optimizer is used to update the weight parameters of the parameterized generator combination according to the minimization of the discriminator loss function, where the generator loss function is expressed by the following formula:

[0076]

[0077] Among them, LG is the generator loss function, LG S is the loss function corresponding to the true or false judgment, i and j are any type of type A, type B and type C corresponding to the preset feature parameters, i′ is the simulation data corresponding to the type represented by i, is the true or false prediction value of the generator in the initial parameterized generator combination recognized by the corresponding discriminator in the initial parameterized discriminator combination, It is the true and false prediction value of the real data generated by the generator in the initial parameterized generator combination, where the true and false prediction value is determined by the sigmoid function, which is expressed by the following formula:

[0078]

[0079] It should be noted that the intelligent seismic synthesis device considering multi-classification characteristics can also add physical information constraints to the response spectrum and energy spectrum during the iteration process, and convert the acceleration response spectra generated before and after the iteration into And real earthquake data For comparison, the details are as follows:

[0080]

[0081] Among them, RMSE is the root mean square error function, They represent the response spectra of the generated earthquake motion and the real earthquake motion, respectively, ε=0.01. At the same time, the response spectrum of the earthquake motion S kIt expresses the relationship between the maximum earthquake acceleration response of a single-degree-of-freedom elastic system and its natural vibration period. Taking the absolute acceleration response spectrum as an example, the specific theoretical formula is as follows:

[0082]

[0083] in, is the ground motion acceleration, ω is the natural frequency of the structure, T is the natural period, ξ is the damping ratio,

[0084] Step A54, fixing the initial parameterized discriminator combination, and determining the weight parameters of the initial parameterized generator combination according to the discriminator loss function corresponding to the initial parameterized discriminator combination;

[0085] In this step, based on the true and false prediction values, the discriminator loss function is calculated, and the weight parameters of the parameterized discriminator combination are updated using the Adam optimizer. Considering the physical constraints of different types of ground motions, the discriminator loss function can be expressed by the following formula:

[0086]

[0087] Wherein, LD is the discriminator loss function, x is any one feature of type A, type B and type C, and p data is the distribution of synthetic seismic data, E is the expected calculated value, and D(x) is the true or false prediction value of the generator output data by the discriminator.

[0088] It should be noted that the intelligent seismic synthesis device considering multi-classification features can also add physical information constraints to the response spectrum and energy spectrum during the process of updating the weight parameters for iteration, and convert the acceleration response spectrum and energy spectrum generated before and after the iteration into And real earthquake data For comparison, the content is:

[0089]

[0090] Among them, RMSE is the root mean square error function, They represent the response spectrum and energy spectrum of the generated ground motion data, They represent the response spectrum and energy spectrum of real earthquake data respectively, ε=0.03.

[0091] Step A55, iteratively execute steps A51 to A54 until a preset iteration termination condition is reached, thereby obtaining a target generator combination and the target discriminator combination.

[0092] In this step, the intelligent seismic synthesis device considering multi-classification features can iteratively execute the above steps until the preset iteration termination condition is reached, and then the optimal weight parameters of the target parameterized generator combination can be obtained. Figure 5 ,After model training, the loss functions of the generator and the discriminator converge rapidly with the ,increase of iteration steps, and the model region is stable.

[0093] See also Figure 6 , Figure 6 A schematic diagram of a certain type of earthquake motion data provided by an embodiment of the present invention generates a certain type of earthquake motion classification label and real earthquake motion data based on a trained target parameterized generator combination to obtain corresponding earthquake motion data.

[0094] refer to Figure 7 The cumulative energy characteristics of the generated seismic waves are consistent with the real ground motion, both of which have obvious inflection points around 20s, and the speed of increase thereafter shows a similar trend; Figure 8 The response spectra of the generated earthquake motions are similar to those of the real earthquake motions, the earthquake characteristic period is about 0.3s, and the response spectra of some generated earthquake motions exceed the standard response spectra, which is consistent with the research results of some literature, that is, the response spectra of some large earthquakes in the sea area will exceed the envelope of the response spectrum of the land area earthquake.

[0095] To sum up, it can be seen that in the embodiments provided in the present application, the seismic motion data, the corresponding classification features and the target generator trained in advance by random noise input are combined to generate corresponding seismic motion synthetic data, fully learn the time domain non-stationarity and multi-classification characteristics of marine seismic motion, and generate seismic motion synthetic data with different characteristic attributes, providing a scientific basis for the simulation of marine engineering earthquake dynamic response and seismic load input.

[0096] The above describes an embodiment of the present invention from the perspective of an intelligent synthesis method of earthquake motion taking into account multiple classification features. The following describes an embodiment of the present invention from the perspective of an intelligent synthesis device of earthquake motion taking into account multiple classification features.

[0097] See also Fig. 9 , a virtual structural diagram of a seismic intelligent synthesis device considering multiple classification features in an embodiment of the present invention, the seismic intelligent synthesis device considering multiple classification features 800 comprises:

[0098] An acquisition module 901 is used to acquire a target three-directional seismic wave data set, where each seismic wave data in the target three-dimensional seismic wave data set includes N data points, where N is an integer greater than or equal to 1;

[0099] A construction module 902 is used to construct a target seismic data matrix based on the target three-directional seismic wave data set;

[0100] The analysis module 903 is used to perform principal component analysis on the target seismic data matrix to obtain target one-dimensional seismic wave data corresponding to each seismic wave data in the target three-directional seismic wave data;

[0101] The generating module 904 is used to splice the target one-dimensional seismic wave data and the random noise corresponding to the target one-dimensional seismic wave data and input them into a pre-trained target generator combination to obtain synthetic seismic motion data.

[0102] In one possible design, the device further includes:

[0103] The training module 905 is used to:

[0104] Acquire m groups of three-dimensional seismic wave data according to the user's operation instruction, wherein each piece of seismic wave data in the m groups of three-dimensional seismic wave data includes n data points, wherein m and n are both integers greater than or equal to 1;

[0105] Constructing a seismic data matrix based on the m groups of three-directional seismic wave data;

[0106] Performing principal component analysis on the seismic data matrix to obtain one-dimensional seismic wave data corresponding to each seismic wave data in the m groups of three-directional seismic wave data;

[0107] The one-dimensional seismic wave data is divided into a first type of seismic wave data, a second type of seismic wave data and a third type of seismic wave data according to preset characteristic parameters, and a labeled seismic motion database is constructed according to the first type of seismic wave data, the second type of seismic wave data and the third type of seismic wave data;

[0108] The initial parameterized generator combination and the initial discriminator combination are trained according to the earthquake database to obtain the target generator combination and the target discriminator combination.

[0109] In a possible design, the training module 905 is specifically used for:

[0110] Performing data standardization processing on each piece of seismic wave data in the target group of three-directional seismic wave data, wherein the target group of three-directional seismic wave data is any one group of three-directional seismic wave data among the m groups of three-directional seismic wave data;

[0111] Calculating the correlation coefficient matrix corresponding to each piece of seismic wave data in the three-directional seismic wave data of the target group;

[0112] The one-dimensional seismic wave data corresponding to each seismic wave data in the three-directional seismic wave data of the target group is determined according to the singular value decomposition method and the correlation coefficient matrix.

[0113] In a possible design, the training module 805 is further specifically used for:

[0114] Inputting the one-dimensional seismic wave data into the initial parameterized generator combination to obtain first simulated seismic wave data and second simulated seismic wave data corresponding to the target seismic wave data;

[0115] Inputting the first simulated seismic wave data, the second simulated seismic wave data and the real one-dimensional seismic wave data into corresponding seismic wave discriminators of various types, and obtaining corresponding true and false prediction values;

[0116] Fixing the initial parameterized generator combination, and determining the loss value of the initial parameterized discriminator combination according to the generator loss function corresponding to the initial parameterized generator combination;

[0117] Fixing the initial parameterized discriminator combination, and determining the weight parameters of the initial parameterized generator combination according to the discriminator loss function corresponding to the initial parameterized discriminator combination;

[0118] The initialized parameterized generator combination and the initialized parameterized discriminator combination are iteratively trained until a preset iteration termination condition is reached, thereby obtaining the target generator combination and the target discriminator combination.

[0119] In one possible design, any parameterized generator in the parameterized generator combination adopts a CNN structure, and the CNN structure includes a forward mapping CNN model and a reverse mapping CNN model. The input of the forward mapping CNN model is one-dimensional seismic wave data, and the input of any parameterized discriminator in the parameterized discriminator combination is the original seismic wave sequence.

[0120] In one possible design, the generator loss function is expressed by the following formula:

[0121]

[0122] Among them, LG is the generator loss function, LG S is the loss function corresponding to the true or false judgment, i and j are respectively any one of the types A, B and C corresponding to the preset feature parameters, i′ is the simulation data corresponding to the type represented by i, is the true or false prediction value of the generator in the initial parameterized generator combination recognized by the corresponding discriminator in the initial parameterized discriminator combination, is the true or false prediction value of the real data generated by the generator in the initial parameterized generator combination.

[0123] In one possible design, the discriminator loss function is expressed by the following formula:

[0124]

[0125] Wherein, LD is the discriminator loss function, x is any one feature of type A, type B and type C, and p data is the distribution of synthetic seismic data, E is the expected calculated value, and D(x) is the true or false prediction value of the generator output data by the discriminator.

[0126] above Fig. 9 The intelligent synthesis device for earthquake motion considering multiple classification features in the embodiment of the present invention is described from the perspective of modular functional entities. The intelligent synthesis device for earthquake motion considering multiple classification features in the embodiment of the present invention is described in detail from the perspective of hardware processing. Fig.10 , a schematic diagram of an embodiment of a seismic intelligent synthesis device 1000 considering multiple classification features in an embodiment of the present invention, the seismic intelligent synthesis device 1000 considering multiple classification features comprises:

[0127] An input device 1001, an output device 1002, a processor 1003 and a memory 1004 (the number of the processor 1003 can be one or more, Fig.10 In some embodiments of the present invention, the input device 1001, the output device 1002, the processor 1003 and the memory 1004 may be connected via a communication bus or other means, wherein: Fig.10 The communication bus connection is taken as an example.

[0128] Wherein, by calling the operation instruction stored in the memory 1004, the processor 1003 is used to perform the following steps:

[0129] Acquire a target three-directional seismic wave data set, wherein each seismic wave data in the target three-dimensional seismic wave data set includes N data points, where N is an integer greater than or equal to 1;

[0130] Constructing a target seismic data matrix based on the target three-directional seismic wave data set;

[0131] Performing principal component analysis on the target seismic data matrix to obtain target one-dimensional seismic wave data corresponding to each seismic wave data in the target three-directional seismic wave data;

[0132] The target one-dimensional seismic wave data, the classification features corresponding to the target one-dimensional seismic wave data, and the random noise corresponding to the target one-dimensional seismic wave data are spliced ​​and input into a pre-trained target generator combination to obtain synthetic seismic motion data.

[0133] By calling the operation instructions stored in the memory 1004, the processor 1003 is also used to execute Figure 1 Any method in the corresponding embodiment.

[0134] See also Fig.11 , Fig.11 A schematic diagram of an electronic device according to an embodiment of the present invention.

[0135] like Fig.11 As shown, an embodiment of the present invention provides an electronic device, including a memory 1110, a processor 1120, and a computer program 1111 stored in the memory 1110 and executable on the processor 1120. When the processor 1120 executes the computer program 1111, the following steps are implemented:

[0136] Acquire a target three-directional seismic wave data set, wherein each seismic wave data in the target three-dimensional seismic wave data set includes N data points, where N is an integer greater than or equal to 1;

[0137] Constructing a target seismic data matrix based on the target three-directional seismic wave data set;

[0138] Performing principal component analysis on the target seismic data matrix to obtain target one-dimensional seismic wave data corresponding to each seismic wave data in the target three-directional seismic wave data;

[0139] The target one-dimensional seismic wave data, the classification features corresponding to the target one-dimensional seismic wave data, and the random noise corresponding to the target one-dimensional seismic wave data are spliced ​​and input into a pre-trained target generator combination to obtain synthetic seismic motion data.

[0140] In a specific implementation process, when the processor 1120 executes the computer program 1111, it can achieve Figure 1 Any implementation manner in the corresponding embodiments.

[0141] Since the electronic device introduced in this embodiment is a device used to implement an intelligent seismic synthesis device that considers multiple classification features in an embodiment of the present invention, based on the method introduced in the embodiment of the present invention, technical personnel in this field can understand the specific implementation mode of the electronic device of this embodiment and its various variations. Therefore, how the electronic device implements the method in the embodiment of the present invention will not be introduced in detail here. As long as the equipment used by technical personnel in this field to implement the method in the embodiment of the present invention is within the scope of protection of the present invention.

[0142] See also Fig.12 , Fig.12 A schematic diagram of an embodiment of a computer-readable storage medium provided in an embodiment of the present invention.

[0143] like Fig.12 As shown, the embodiment of the present invention further provides a computer-readable storage medium 1200, on which a computer program 1211 is stored. When the computer program 1211 is executed by a processor, the following steps are implemented:

[0144] Acquire a target three-directional seismic wave data set, wherein each seismic wave data in the target three-dimensional seismic wave data set includes N data points, where N is an integer greater than or equal to 1;

[0145] Constructing a target seismic data matrix based on the target three-directional seismic wave data set;

[0146] Performing principal component analysis on the target seismic data matrix to obtain target one-dimensional seismic wave data corresponding to each seismic wave data in the target three-directional seismic wave data;

[0147] The target one-dimensional seismic wave data, the classification features corresponding to the target one-dimensional seismic wave data, and the random noise corresponding to the target one-dimensional seismic wave data are spliced ​​and input into a pre-trained target generator combination to obtain synthetic seismic motion data.

[0148] In the specific implementation process, the computer program 1211 is executed by the processor to achieve Figure 1 Any implementation manner in the corresponding embodiments.

[0149] It should be noted that in the above embodiments, the description of each embodiment has its own emphasis, and for parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0150] The embodiment of the present invention also provides a computer program product, which includes computer software instructions. When the computer software instructions are executed on a processing device, the processing device executes the following Figure 1 The process in the corresponding embodiment.

[0151] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present invention is fully or partially generated.

[0152] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0153] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for intelligent synthesis of ground motions considering multi-classification features, characterized in that: include: Acquire a target three-directional seismic wave data set, wherein each seismic wave data in the target three-dimensional seismic wave data set includes N data points, where N is an integer greater than or equal to 1; Constructing a target seismic data matrix based on the target three-directional seismic wave data set; Performing principal component analysis on the target seismic data matrix to obtain target one-dimensional seismic wave data corresponding to each seismic wave data in the target three-directional seismic wave data; The target one-dimensional seismic wave data and the random noise corresponding to the target one-dimensional seismic wave data are spliced ​​and input into a pre-trained target generator combination to obtain synthetic seismic motion data.

2. The method according to claim 1, characterized in that The method further comprises: Acquire m groups of three-dimensional seismic wave data according to the user's operation instruction, wherein each piece of seismic wave data in the m groups of three-dimensional seismic wave data includes n data points, wherein m and n are both integers greater than or equal to 1; Constructing a seismic data matrix based on the m groups of three-directional seismic wave data; Performing principal component analysis on the seismic data matrix to obtain one-dimensional seismic wave data corresponding to each seismic wave data in the m groups of three-directional seismic wave data; The one-dimensional seismic wave data is divided into a first type of seismic wave data, a second type of seismic wave data and a third type of seismic wave data according to preset characteristic parameters, and a labeled seismic motion database is constructed according to the first type of seismic wave data, the second type of seismic wave data and the third type of seismic wave data; The initial parameterized generator combination and the initial discriminator combination are trained according to the earthquake database to obtain the target generator combination and the target discriminator combination.

3. The method according to claim 2, characterized in that The performing principal component analysis on the seismic data matrix to obtain one-dimensional seismic wave data comprises: Performing data standardization processing on each piece of seismic wave data in the target group of three-directional seismic wave data, wherein the target group of three-directional seismic wave data is any one group of three-directional seismic wave data among the m groups of three-directional seismic wave data; Calculating the correlation coefficient matrix corresponding to each piece of seismic wave data in the three-directional seismic wave data of the target group; The one-dimensional seismic wave data corresponding to each seismic wave data in the three-directional seismic wave data of the target group is determined according to the singular value decomposition method and the correlation coefficient matrix.

4. The method according to claim 2, characterized in that: The training of the initial parameterized generator combination and the initial parameterized discriminator according to the earthquake database to obtain the target generator combination and the target discriminator combination includes: Inputting the one-dimensional seismic wave data into the initial parameterized generator combination to obtain first simulated seismic wave data and second simulated seismic wave data corresponding to the target seismic wave data; Inputting the first simulated seismic wave data, the second simulated seismic wave data and the real one-dimensional seismic wave data into corresponding seismic wave discriminators of various types, and obtaining corresponding true and false prediction values; Fixing the initial parameterized generator combination, and determining the loss value of the initial parameterized discriminator combination according to the generator loss function corresponding to the initial parameterized generator combination; Fixing the initial parameterized discriminator combination, and determining the weight parameters of the initial parameterized generator combination according to the discriminator loss function corresponding to the initial parameterized discriminator combination; The initialized parameterized generator combination and the initialized parameterized discriminator combination are iteratively trained until a preset iteration termination condition is reached, thereby obtaining the target generator combination and the target discriminator combination.

5. The method according to claim 4, characterized in that Any parameterized generator in the parameterized generator combination adopts a CNN structure, and the CNN structure includes a forward mapping CNN model and a reverse mapping CNN model. The input of the forward mapping CNN model is one-dimensional seismic wave data, and the input of any parameterized discriminator in the parameterized discriminator combination is an original seismic wave sequence.

6. The method according to claim 4, characterized in that The generator loss function is expressed by the following formula: Among them, LG is the generator loss function, LG S is the loss function corresponding to the true or false judgment, i and j are respectively any one of the types A, B and C corresponding to the preset feature parameters, i′ is the simulation data corresponding to the type represented by i, is the true or false prediction value of the generator in the initial parameterized generator combination recognized by the corresponding discriminator in the initial parameterized discriminator combination, is the true or false prediction value of the real data generated by the generator in the initial parameterized generator combination.

7. The method according to claim 4, characterized in that The discriminator loss function is expressed by the following formula: Wherein, LD is the discriminator loss function, x is any one feature of type A, type B and type C, and p data is the distribution of synthetic seismic data, E is the expected calculated value, and D(x) is the true or false prediction value of the generator output data by the discriminator.

8. An intelligent synthesis device for earthquake motion considering multi-classification features, characterized in that: include: An acquisition module, used for acquiring a target three-directional seismic wave data set, wherein each seismic wave data in the target three-dimensional seismic wave data set includes N data points, where N is an integer greater than or equal to 1; A construction module, used for constructing a target seismic data matrix based on the target three-directional seismic wave data set; An analysis module, used for performing principal component analysis on the target seismic data matrix to obtain target one-dimensional seismic wave data corresponding to each seismic wave data in the target three-directional seismic wave data; The generation module is used to splice the target one-dimensional seismic wave data and the random noise corresponding to the target one-dimensional seismic wave data and input them into a pre-trained target generator combination to obtain seismic synthetic data.

9. An electronic device, characterized in that: include: A memory and a processor, wherein the processor is used to implement the steps of the intelligent synthesis method of earthquake motion considering multiple classification features as described in any one of claims 1 to 7 when executing the computer management program stored in the memory.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer management program, which, when executed by a processor, implements the steps of the intelligent synthesis method for earthquake motion considering multiple classification features as described in any one of claims 1 to 7.

Images