An analysis method and system based on seismic wave images

Abstract

The present invention provides an analysis method and system based on seismic wave images, which relate to the field of earthquake early warning technology, including obtaining seismic wave image data. Sending it to a feature extraction model for processing, extracting spatiotemporal feature data. Using multi-scale adaptive bandpass filtering to analyze the spatiotemporal feature data, obtain the initial estimated position information of the earthquake source. Based on this information, the historical response of the corresponding building structure is determined, and a nonlinear response model is obtained by training through a deep autoencoder network. The fractal dimension of the seismic wave data is calculated using a multi-fractal analysis method, a spatial influence diagram of the earthquake source is generated, the dominant frequency characteristics of the earthquake source are extracted, and a response prediction model is constructed in combination with the historical response information of the building structure. The response of the building structure is predicted in real time by the model, and earthquake damage warning is performed based on the prediction results. The present invention can improve the accuracy and efficiency of emergency response to earthquake disasters and enhance the safety of buildings.

Description

Technical Field

[0001] The present invention relates to the field of earthquake early warning technology, and in particular to an analysis method and system based on seismic wave images. Background Art

[0002] In the field of earthquake prediction and building structure safety analysis, the relationship between the propagation characteristics of seismic waves and the dynamic response of building structures has attracted much attention. In the prior art, seismic wave monitoring mainly relies on single signal data such as acceleration or velocity, and usually uses conventional frequency domain analysis methods such as bandpass filtering or Fourier transform to extract basic frequency characteristics of seismic waves. However, these traditional methods often find it difficult to obtain detailed spatial location of the earthquake source and dominant frequency characteristics when faced with the complex nonlinear characteristics of seismic waves. At the same time, the historical response data of building structures are mostly processed using linear analysis models, which limits the accuracy of structural response prediction. Especially in the case of complex seismic waves, traditional methods are difficult to accurately predict the response of building structures in future earthquakes. In addition, the earthquake source location and building structure response analysis models in the prior art are mostly based on fixed filters or single-scale analysis, which makes it difficult to dynamically adapt to changes in seismic wave characteristics at different scales, affecting the accuracy of earthquake source location and the effectiveness of structural response prediction.

[0003] Therefore, there is an urgent need for an analysis method and system based on seismic wave images to improve the accuracy of earthquake source location and the effectiveness of structural response prediction. Summary of the invention

[0004] The purpose of the present invention is to provide an analysis method and system based on seismic wave images to improve the above problems. In order to achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0005] In a first aspect, the present invention provides an analysis method based on seismic wave images, comprising:

[0006] Acquire seismic wave image data, wherein the seismic wave image data includes acceleration data, displacement data, velocity data, timestamp data and direction data of the seismic wave;

[0007] Sending the seismic wave image data to a preset feature extraction model for processing to obtain spatiotemporal feature data of the seismic wave image data;

[0008] Analyzing the spatiotemporal characteristic data of the seismic wave image data based on a preset multi-scale adaptive bandpass filter to obtain initial estimated position information of the earthquake source corresponding to the seismic wave image data;

[0009] Determine the historical response information of the corresponding building structure based on the initial estimated position information of the earthquake source, and send the initial estimated position information of the earthquake source and the historical response information of the corresponding building structure to a preset deep autoencoder network for training to obtain a nonlinear response model;

[0010] Based on the nonlinear response model and a preset multi-fractal analysis method, the fractal dimension of the seismic wave image data is calculated to obtain the influence intensity of the earthquake source at different spatial positions, and a spatial influence diagram of the earthquake source is generated based on the influence intensity of the earthquake source at different spatial positions;

[0011] Extracting the dominant frequency characteristics of the earthquake source based on the spatial influence diagram of the earthquake source, and constructing a building structure response prediction model based on the extracted dominant frequency characteristics of the earthquake source and the historical response information of the building structure;

[0012] The building structure response is predicted in real time based on the building structure response prediction model, and a real-time warning is given based on the prediction result.

[0013] In a second aspect, the present invention further provides an analysis system based on seismic wave images, comprising:

[0014] An acquisition unit, used for acquiring seismic wave image data, wherein the seismic wave image data includes acceleration data, displacement data, velocity data, time stamp data and direction data of the seismic wave;

[0015] A processing unit, used for sending the seismic wave image data to a preset feature extraction model for processing to obtain spatiotemporal feature data of the seismic wave image data;

[0016] An analysis unit, configured to analyze the spatiotemporal characteristic data of the seismic wave image data based on a preset multi-scale adaptive bandpass filter to obtain initial estimated position information of a seismic source corresponding to the seismic wave image data;

[0017] A training unit, used to determine the historical response information of the corresponding building structure based on the initial estimated position information of the earthquake source, and send the initial estimated position information of the earthquake source and the historical response information of the corresponding building structure to a preset deep autoencoder network for training to obtain a nonlinear response model;

[0018] A calculation unit, configured to calculate the fractal dimension of the seismic wave image data based on the nonlinear response model and a preset multi-fractal analysis method, obtain the influence intensity of the seismic source at different spatial positions, and generate a spatial influence diagram of the seismic source based on the influence intensity of the seismic source at different spatial positions;

[0019] An extraction unit, used for extracting the dominant frequency characteristics of the earthquake source based on the spatial influence diagram of the earthquake source, and constructing a building structure response prediction model based on the extracted dominant frequency characteristics of the earthquake source and the historical response information of the building structure;

[0020] The early warning unit is used to predict the building structure response in real time based on the building structure response prediction model, and to issue a real-time early warning based on the prediction result.

[0021] The beneficial effects of the present invention are:

[0022] The present invention proposes an analysis method and system based on seismic wave images, which extracts the nonlinear spatiotemporal characteristics of seismic waves by integrating multi-scale adaptive bandpass filtering and deep autoencoder networks, and conducts comprehensive analysis in combination with the historical response information of building structures. This method optimizes the feature extraction of seismic wave images through manifold learning and t-SNE dimensionality reduction technology, making the initial estimated position of the earthquake source more accurate; it further extracts the dominant frequency characteristics of the earthquake source by combining wavelet packet decomposition and synchronous phase analysis, realizes the effective correlation between the frequency characteristics of the earthquake source and the response characteristics of the building structure, and forms a structural response prediction model. Compared with traditional methods, the present invention has significantly improved the accuracy of earthquake source location and structural response prediction, and can support real-time earthquake warning and enhance the effectiveness of building structure safety assessment.

[0023] Other features and advantages of the present invention will be set forth in the following description, and in part will become apparent from the description, or may be understood by practicing embodiments of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0025] Figure 1 Schematic diagram of the flow of the analysis method based on seismic wave images described in an embodiment of the present invention;

[0026] Figure 2 Schematic diagram of the structure of the seismic wave image-based analysis system described in an embodiment of the present invention.

[0027] In the figure: 701, acquisition unit; 702, processing unit; 703, analysis unit; 704, training unit; 705, calculation unit; 706, extraction unit; 707, early warning unit. DETAILED DESCRIPTION

[0028] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0029] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0030] Embodiment 1:

[0031] This embodiment provides an analysis method based on seismic wave images.

[0032] See also Figure 1 , the figure shows that the method includes step S1, step S2, step S3, step S4, step S5, step S6 and step S7.

[0033] Step S1, obtaining seismic wave image data, wherein the seismic wave image data includes acceleration data, displacement data, velocity data, timestamp data and direction data of the seismic wave;

[0034] It is understandable that obtaining seismic wave image data is one of the basic steps. These data mainly include information such as acceleration, displacement, velocity, timestamp and direction of seismic waves. Each data type captures the dynamic characteristics of seismic waves from different angles. Acceleration data is used to analyze the instantaneous energy changes of seismic waves and is an important parameter for judging the intensity and impact of seismic waves; displacement and velocity data reflect the displacement amplitude and propagation velocity during the propagation of seismic waves, thereby helping to identify the wave characteristics and energy attenuation pattern of the source. Timestamp data accurately locates seismic waves in time series, laying the foundation for time-frequency analysis; direction data can indicate the main direction of seismic wave propagation, providing a spatial reference for subsequent source location and impact range determination. The acquisition of this multidimensional data set can more comprehensively reflect the characteristics of seismic waves and meet the needs of subsequent multi-scale analysis. In the current application, this comprehensive set of seismic wave data greatly improves the accuracy and completeness of source information extraction, provides comprehensive input support for deep autoencoder networks and multi-scale analysis algorithms, and can improve the accuracy of source location and frequency feature extraction in subsequent processing.

[0035] Step S2, sending the seismic wave image data to a preset feature extraction model for processing to obtain spatiotemporal feature data of the seismic wave image data;

[0036] It can be understood that this step is based on the algorithm of manifold learning, and uses the neighborhood relationship between data points to construct a high-dimensional manifold model of seismic wave data, thereby revealing the potential nonlinear structure and spatiotemporal variation law in the data. In addition, using dimensionality reduction techniques such as t-SNE to map high-dimensional manifold data to low-dimensional space can simplify the data structure and reveal the potential patterns and trends therein, providing a clearer spatiotemporal feature map for subsequent analysis. In this step, step S2 includes step S21, step S22 and step S23.

[0037] Step S21, modeling the seismic wave image data based on a preset manifold learning method, wherein the domain relationship of each data point in the seismic wave image data and the distance between the data points are defined, and a manifold model is constructed based on the selected neighborhood relationship and the gradient information of each data point on the manifold model is calculated;

[0038] It can be understood that this step first defines the domain relationship of each seismic wave data point and the distance between data points. This step is to determine which data points can be regarded as each other's "neighbors" to ensure that the model can truly reflect the local correlation of data points in space. Then, based on the selected neighborhood relationship, a manifold model is constructed to distribute and represent high-dimensional seismic wave data on the manifold structure. In this process, the role of the manifold learning method is to maintain the local geometric information in the high-dimensional data and map it to the model for subsequent dimensionality reduction analysis or feature extraction.

[0039] It can be understood that this step of analysis based on manifold learning can fully tap the nonlinear characteristics of seismic wave data, laying the foundation for extracting spatiotemporal feature maps in subsequent steps. In traditional methods, direct processing of seismic wave data is prone to missing nonlinear correlations, while manifold learning optimizes data representation by maintaining neighborhood relationships and local geometric features, which helps to better capture the complex changes in space and time of seismic waves. The calculation formula for gradient information is as follows:

[0040]

[0041] Among them, g i Represents the gradient information of the data point, W ij represents the distance weight matrix in the neighborhood, x j represents the jth data point, μ i represents the local average vector in the neighborhood, and k represents the total number of data points.

[0042] Step S22, mapping the gradient information of each data point on the manifold model to a low-dimensional space for dimensionality reduction based on a preset t-SNE algorithm to obtain a spatiotemporal feature map of the seismic wave image data;

[0043] It is understandable that the t-SNE algorithm is a nonlinear dimensionality reduction method that focuses on preserving the neighborhood structure of data points in the original high-dimensional space. It can effectively show the relationship between neighboring points in high-dimensional data. Here, the t-SNE algorithm first receives the gradient information after manifold learning processing and converts it into a low-dimensional representation. The core of dimensionality reduction is to compress complex seismic wave data from high dimensions to two-dimensional or three-dimensional space, so that the spatiotemporal features can be more clearly and intuitively displayed and reduce the consumption of computing resources. Moreover, the t-SNE algorithm can well capture the local similarity of data and retain the neighboring points in the original high-dimensional space in the low-dimensional space. This method shows the spatial pattern and dynamic characteristics of seismic wave data in the spatiotemporal feature map. The generated spatiotemporal feature map not only contains the local correlation characteristics of seismic waves, but also retains the changing trend during the propagation of seismic waves, providing structured information input for subsequent feature analysis and modeling.

[0044] Step S23: extracting and processing the spatiotemporal characteristic data from the spatiotemporal characteristic graph of the seismic wave image data based on the principal component analysis method to obtain spatiotemporal characteristic data.

[0045] It can be understood that this step first centralizes the data and then calculates the covariance matrix of the data to reveal the correlation structure between different features. Next, the main components (i.e., principal components) of the data set are found by solving the eigenvalues ​​and eigenvectors of the covariance matrix. These principal components represent the main directions of data change. Generally, retaining the principal components with larger eigenvalues ​​can effectively reduce the data dimension because these principal components carry most of the information of the original data. This step can improve the computational efficiency of the system while enhancing the generalization ability of the prediction model.

[0046] Step S3, analyzing the spatiotemporal characteristic data of the seismic wave image data based on a preset multi-scale adaptive bandpass filter to obtain initial estimated position information of the earthquake source corresponding to the seismic wave image data;

[0047] It is understandable that in this step, the temporal and spatial characteristic data of the seismic wave image are decomposed in the frequency domain by a preset multi-scale adaptive bandpass filter to extract the initial position information of the source. The multi-scale adaptive bandpass filter extracts the specific frequency band characteristics of the seismic wave signal layer by layer within different frequency ranges, and effectively separates the main frequency components in the source signal. The purpose of doing this is to identify and locate the source according to the frequency difference of different seismic wave characteristics. The multi-scale frequency division analysis enhances the ability to capture the dominant frequency and propagation characteristics of the seismic wave, making the initial location of the source more accurate, and effectively filtering out high-frequency noise and low-frequency background signals. This method can significantly improve the response sensitivity of the system, especially when the seismic signal is relatively weak or the interference is large, the source position can be estimated more accurately, and key position information can be provided for subsequent building structure response analysis. Among them, the multi-scale adaptive bandpass filtering combines the advantages of multi-scale analysis and adaptive filters, making signal processing more flexible and accurate, and being able to capture important features in different frequency bands and different time and space scales in seismic waves. Among them, multi-scale analysis refers to processing a signal at multiple different scales (or frequencies), and adaptive filtering refers to that the parameters of the filter (such as frequency and bandwidth) can be dynamically adjusted according to the characteristics of the input signal to adapt to changes in the signal. In this step, step S3 includes step S31, step S32, step S33 and step S34.

[0048] Step S31, constructing a bandpass filter based on the spatiotemporal feature data, wherein the center frequency and bandwidth of each bandpass filter are set by a window function to obtain at least two bandpass filters of different frequency bands;

[0049] It can be understood that this step separates the different frequency band information in the seismic signal through the filter in order to better extract the source location characteristics. In the filter construction, the key parameters of the bandpass filter, the center frequency and bandwidth, are first determined. The setting of these parameters is based on the frequency distribution of the seismic wave and the characteristic frequency of the source information. We define these frequencies and bandwidths through window functions, which can effectively control the range of the frequency band so that the filter can focus on processing the signal characteristics within the target frequency range. The multi-band filter setting allows for detailed source feature extraction over a wider frequency range, enabling the system to adapt to changes in seismic waves at different scales, thereby improving the accuracy of the initial location estimation of the source. At the same time, this multi-scale, multi-band filtering analysis strategy has good adaptability in signal processing, and can flexibly adjust the bandwidth according to the characteristics of different seismic wave signals to retain the target features to the maximum extent.

[0050] Step S32, filtering the spatiotemporal feature data based on the bandpass filters of the different frequency bands, wherein all the bandpass filters are applied to the image data in the spatiotemporal feature data through convolution operation to obtain the image data of each frequency band after bandpass filtering;

[0051] It is understandable that the core function of the bandpass filter is to retain the signal characteristics within a certain frequency band while suppressing low-frequency and high-frequency signals. In this step, bandpass filters of different frequency bands are used to extract different frequency components in the spatiotemporal feature data respectively, so that different information in the seismic wave image data can be processed in a targeted manner. Each bandpass filter has a unique center frequency and bandwidth. Through these filters, the frequency components of the signal will be separated into multiple different frequency bands. Thus, this step can effectively extract the characteristics of the seismic wave signal in each frequency range, help eliminate noise and enhance the source and wave information, and provide accurate frequency band data support for subsequent earthquake source location and source analysis.

[0052] Step S33, calculating the local maximum value based on the image data of each frequency band after the bandpass filtering, and comparing the local maximum value with a preset threshold value to obtain a comparison result;

[0053] It can be understood that the local maximum in this step refers to a specific position, and the value in its neighborhood is greater than the values ​​of all surrounding adjacent positions. In image data processing, local maximum is usually used to identify significant features, fluctuation peaks or key events in the image. In this step, the image data of each frequency band after bandpass filtering has been processed by filters of different frequency bands, so as to extract information of different frequency components. The significant fluctuations in these image data are found by calculating the local maximum. Among them, by calculating the local maximum and performing threshold comparison, the important source features in the image can be accurately extracted, which helps to determine the source location or epicenter.

[0054] Step S34: based on the comparison result, at least two candidate earthquake source positions are determined, and all the candidate earthquake source positions are weightedly calculated to obtain an initial estimated earthquake source position corresponding to the seismic wave image data.

[0055] It can be understood that this step can combine the advantages and disadvantages of different candidate locations by weighted calculation of multiple candidate source locations, thereby improving the accuracy of the initial estimated location of the source. In the presence of noise or incomplete signals, the weighted calculation of candidate source locations can effectively reduce errors and improve the robustness of seismic wave image data analysis.

[0056] The weighted calculation formula is as follows:

[0057]

[0058] Among them, Q is the final initial estimated location of the earthquake source, d represents the index of each candidate earthquake source location, and v d represents the weight corresponding to each candidate source location, z d represents the specific location information of the candidate location, and r represents the total number of candidate source locations.

[0059] Step S4, determining the historical response information of the corresponding building structure based on the initial estimated position information of the earthquake source, and sending the initial estimated position information of the earthquake source and the historical response information of the corresponding building structure to a preset deep autoencoder network for training to obtain a nonlinear response model;

[0060] It is understandable that this step determines the historical response information of the building structure associated with the source location based on the initial estimated location information of the earthquake source. The historical response of the building structure usually includes the response data of the structure such as vibration, displacement, acceleration, etc. in different time periods, which reflects the actual response of the building during the propagation of seismic waves. The deep autoencoder network is an unsupervised learning model that extracts important features from the data by encoding and decoding the input data. The encoder part of the network maps the historical response information and the source location into a low-dimensional latent space in order to better capture the inherent correlation of these complex data, while the decoder reconstructs the low-dimensional representation into the original data. In this way, the training of the deep learning model automatically extracts the nonlinear mapping relationship between the source location and the historical response of the building structure, and then establishes an accurate nonlinear response model. It is understandable that in this step, step S4 includes step S41, step S42 and step S43.

[0061] Step S41, based on the above-mentioned initial estimated position information of the earthquake source, determine the propagation path from the initial estimated position of the earthquake source to the building structure at the corresponding position, and compare the propagation path of the initial estimated position of the earthquake source with the propagation path in the historical records through a similarity matching algorithm to obtain the historical response information of the building structure corresponding to the initial estimated position of the earthquake source;

[0062] It can be understood that this step determines the propagation path from the earthquake source position to the building structure through the initial estimated position information of the earthquake source. This propagation path represents the process of earthquake waves propagating from the earthquake source to the building. Taking into account the geometric characteristics, material properties and propagation characteristics of the building structure, the determination of the propagation path in this step usually depends on the earthquake wave equation. The propagation path of the initial estimated position of the earthquake source is compared with the propagation path in the historical records through the similarity matching algorithm. The propagation path in the historical records is usually collected through previous earthquake events and building response data. These data provide a real historical trajectory for analyzing the interaction between the earthquake source and the building. The core function of the similarity matching algorithm is to calculate the similarity between the two paths. The cosine similarity calculation method is used in this step.

[0063] Step S42, extracting the time-frequency characteristics of the seismic wave image data based on the segmented Fourier transform, and decomposing the historical response information of the building structure corresponding to the initial estimated position of the earthquake source based on the wavelet transform to obtain the structural response characteristic data;

[0064] It can be understood that this step first extracts the time-frequency characteristics of seismic wave image data based on the segmented Fourier transform. The segmented Fourier transform is a commonly used technique for analyzing the time-frequency characteristics of signals. It divides the signal into several time periods and applies the Fourier transform to each time period respectively, thereby obtaining frequency domain information in different time windows. This method helps to identify the changing characteristics of seismic wave signals in time and frequency, especially for the analysis of non-stationary signals (such as seismic waves). Through the segmented Fourier transform, the frequency components of the seismic wave signal can be captured, and then the historical response information of the building structure corresponding to the initial estimated position of the source is decomposed based on the wavelet transform. Wavelet transform is a time-frequency localization transform method. Compared with Fourier transform, it can provide information in both the time domain and the frequency domain. Through wavelet transform, the historical response information of the building structure is decomposed into components of different frequencies while maintaining the locality in time. Wavelet transform is particularly suitable for analyzing irregular and instantaneous changing signals because it can capture the local characteristics of the signal at different scales, such as sudden changes in building response, vibration patterns, and long-term vibration trends, thereby combining the advantages of segmented Fourier transform and wavelet transform. The characteristics of earthquake waves and building responses are deeply explored in terms of both time and frequency. Through this multi-level analysis, more detailed and accurate response predictions can be provided for building seismic design, thereby improving the accuracy of building seismic capacity assessment and disaster response strategies.

[0065] Step S43: sending the time-frequency feature and the structural response feature data to a preset autoencoder network for training, wherein a mean square error function is used as a loss function, and the model parameters are updated by a stochastic gradient descent algorithm to obtain a trained nonlinear response model.

[0066] It can be understood that the autoencoder is an unsupervised learning model, which is often used for feature learning and data dimensionality reduction. Its structure includes an encoder and a decoder. The encoder part maps the input data to a low-dimensional latent space, while the decoder reconstructs the input data from the latent space. Through the training process, the autoencoder can learn the potential structural features in the input data. Through the training of the autoencoder, useful features can be extracted from complex seismic waves and structural response data, and the model can be optimized by the stochastic gradient descent algorithm so that it can accurately capture the nonlinear response of the building structure in an earthquake. This method greatly improves the accuracy of seismic response analysis and can provide a scientific basis for the seismic optimization of buildings.

[0067] Step S5, performing fractal dimension calculation on the seismic wave image data based on the nonlinear response model and a preset multi-fractal analysis method to obtain the influence intensity of the earthquake source at different spatial positions, and generating a spatial influence diagram of the earthquake source based on the influence intensity of the earthquake source at different spatial positions;

[0068] It can be understood that the multifractal analysis method is a technique for analyzing signals with complex structures, scale invariance or self-similarity. Seismic wave image data usually has the characteristics of multi-scale and nonlinearity, and is suitable for using multifractal theory to describe its complexity. In multifractal analysis, fractal dimensions are usually calculated, that is, the degree of self-similarity of the signal or image at different scales is described. The multifractal analysis method can provide a multi-scale description for seismic wave image data, thereby more accurately revealing the influence of the source at different positions. The specific operation includes selecting an appropriate scale interval and performing multi-scale decomposition of the image data. By analyzing the local structure at different scales, the local fractal dimension at the scale is calculated. By cumulative calculation, the overall fractal dimension distribution of the image at different scales is obtained to form a multifractal spectrum. In this step, step S5 includes step S51, step S52, step S53, step S54 and step S54.

[0069] Step S51, dividing the seismic wave image data evenly according to the initial estimated location of the earthquake source to obtain at least two seismic wave sub-images;

[0070] It is understandable that this step selects the initial estimated position of the earthquake source as the center point of segmentation, and symmetrically segments the image on the left and right, and on the top and bottom of the earthquake source. The segmentation method can adopt rectangular segmentation. Through the segmentation of the initial estimated position of the earthquake source, multiple local sub-images can be obtained. Each of these sub-images contains a part of the seismic wave image, focusing on the more important area near the earthquake source or in the direction of the earthquake source. The data in each seismic wave sub-image can reflect the seismic wave propagation characteristics of the area, including information such as the intensity, propagation direction and attenuation of the seismic waves emitted by the earthquake source.

[0071] Step S52, calculating the entropy change rate of all seismic wave sub-images to obtain the entropy change rate corresponding to each seismic wave sub-image;

[0072] It is understandable that the calculation of the entropy change rate in this step usually requires first calculating the entropy value of each sub-image. The entropy change rate directly reflects the propagation characteristics of seismic waves, especially the complexity difference between the source area and other areas. If the entropy change rate in some areas is high, it may indicate that the propagation of seismic waves in these areas is more complex, which may be strongly affected by the source or there are more reflection and diffraction effects. Among them, the entropy value calculation formula and entropy change rate calculation formula for seismic wave sub-images are as follows:

[0073]

[0074] in, is the entropy value, indicating that the data set is the amount of information or uncertainty, m is the number of different possible values ​​in the data set, and zn is the nth value in the data set, p(z n ) is z n The probability distribution of .

[0075]

[0076] Among them, D H is the dimension of entropy, N(∈) represents the number of boxes required to cover the data set under the box size ∈, and ∈ is the box size.

[0077] Step S53, determining structural response characteristic data of each seismic wave sub-image based on the nonlinear response model, and then obtaining intensity distribution data of different sub-regions;

[0078] It can be understood that this step calculates the structural response characteristic data of each seismic wave sub-image based on the nonlinear response model, thereby obtaining the intensity distribution data of different sub-regions. The nonlinear response model is obtained by modeling the historical response information and seismic wave data through deep learning and autoencoder technology in the previous training process. This model can capture the complex nonlinear relationship between seismic waves and building structures, especially effectively modeling the nonlinear response of the structure, showing the interaction between the source and the building under the actual earthquake. The structural response characteristic data of each seismic wave sub-image is obtained by inputting the seismic wave image into the nonlinear response model. The model predicts the response of the structure at this position based on the different initial positions of the source and the corresponding historical response data. These structural response characteristic data usually include dynamic response indicators such as displacement, acceleration, and velocity of the building structure, which can reflect the deformation and vibration mode of the structure under the action of seismic waves.

[0079] Through this step, the intensity distribution data can accurately reveal the impact of earthquake waves on different areas. Through this step, technicians can gain a deep understanding of the intensity distribution of earthquake waves at different locations of the building structure, thereby providing an important basis for the seismic design of the building. For example, in high-risk areas, building structures may show stronger nonlinear responses, resulting in excessive deformation or damage in local areas. This intensity distribution data can provide data support for further optimizing seismic structure design and emergency response.

[0080] Step S54, performing fractal dimension calculation on the intensity distribution data of all sub-regions to obtain the local fractal dimension of each sub-region;

[0081] It can be understood that this step uses fractal analysis methods to quantify the complexity of each area in the image. The larger the fractal dimension, the more complex the reaction characteristics of the area. Conversely, the lower the fractal dimension, the simpler the reaction characteristics of the area. Generally, the calculation of fractal dimension can be achieved by fractal box counting. These methods divide each sub-area of ​​the image into multiple small blocks and determine its dimension value by calculating the intensity distribution change in each small block.

[0082] The calculation formula of the local fractal dimension is as follows:

[0083]

[0084] Where D represents the local fractal dimension, δ represents the subregion size, and N(δ) represents the number of non-zero intensity distribution blocks in the grid.

[0085] Step S55: perform spatial interpolation based on the local fractal dimensions of all sub-regions, and normalize the fractal dimension of each sub-region obtained by interpolation, thereby generating a spatial influence map of the earthquake source.

[0086] It is understandable that this step, by the Kriging interpolation method, extends the discrete local fractal dimension data smoothly to the entire spatial region. The purpose of doing so is to make up for the fractal dimension value difference between each sub-region, so that the fractal dimension distribution in the entire space is smoother and more coherent, which is convenient for subsequent analysis. After completing the interpolation, the interpolation fractal dimension of each sub-region is normalized using the minimum-maximum normalization method, that is, all interpolation values ​​are mapped to an interval of 0 to 1. Such processing helps to unify the scales of different regions, making the spatial distribution of the intensity of the source impact more intuitive, and is convenient for subsequent image analysis and the comprehensive assessment of the source position and impact. This spatial influence diagram can effectively present the overall impact range of the source on the structure, clearly showing the degree of impact in different regions, and is convenient for predicting and analyzing the impact range and intensity changes of seismic waves.

[0087] Step S6, extracting the dominant frequency characteristics of the earthquake source based on the spatial influence diagram of the earthquake source, and constructing a building structure response prediction model based on the extracted dominant frequency characteristics of the earthquake source and the historical response information of the building structure;

[0088] It can be understood that this step uses the spatial influence diagram of the earthquake source to extract the dominant frequency characteristics of the earthquake source, thereby constructing a more accurate building structure response prediction model. The accuracy of the prediction model is improved, so that the building structure can make a more reasonable response assessment for different earthquake source frequencies and amplitude characteristics, thereby facilitating emergency response to earthquake disasters and improving building safety. In this step, step S6 includes step S61, step S62, step S63 and step S64.

[0089] Step S61, performing wavelet packet decomposition on the spatial influence diagram of the earthquake source to obtain earthquake source time evolution data of at least two frequency bands;

[0090] It can be understood that by performing wavelet packet decomposition on the spatial influence map of the earthquake source, the time evolution data of the earthquake source in different frequency bands can be obtained. Wavelet packet decomposition can not only provide local information of time and frequency, but also finely decompose the high-frequency and low-frequency components to extract the characteristics of the earthquake source signal in different time and frequency bands. Wavelet packet decomposition is a process of recursively decomposing signals. Its basic idea is to further decompose the low-frequency and high-frequency components of the signal on the basis of traditional wavelet decomposition to construct a more complete frequency band subdivision. First, the original signal is decomposed into two sub-bands of low frequency and high frequency through a wavelet filter bank; then, the obtained low-frequency and high-frequency sub-bands are further decomposed to obtain higher-level refined frequency bands. This process is continuously recursive to generate a decomposition tree consisting of multiple layers of sub-bands. Through wavelet packet decomposition, the signal is divided into equal-width frequency bands. This refined structure provides more accurate frequency resolution, so that signal features in different frequency ranges can be extracted, so that the details of complex signals can be captured simultaneously in the time domain and frequency domain.

[0091] Step S62, identifying the phase synchronization of the earthquake source time evolution data of all frequency bands by synchronous phase analysis method, and obtaining the phase synchronization characteristics of the earthquake source;

[0092] It can be understood that this step uses the synchronous phase analysis method to identify the phase synchronization of the time evolution data of the source in each frequency band to obtain the phase synchronization characteristics of the source. Phase synchronization analysis is a method for identifying the phase relationship between signals in the frequency or time-frequency domain. First, the time evolution data of each frequency band is preprocessed to ensure that it is suitable for phase analysis; then, the signal is converted to the phase space by Hilbert transform to identify the change pattern of the phase angle. The degree of phase synchronization is determined by calculating the phase difference and consistency between different frequency bands. If the phase relationship between frequency bands is highly consistent, it indicates that the source signals between the frequency bands have a high degree of synchronization. This phase synchronization feature can reveal the coordinated behavior of the source in different frequency bands, which helps to more deeply understand the propagation characteristics of the source signal and the coupling between frequencies. The phase synchronization feature extracted in this step can provide information on the consistency of the source frequency, providing an accurate basis for the prediction of the source influence range in the subsequent model.

[0093] Step S63, performing cluster analysis based on the phase synchronization characteristics of the earthquake sources to obtain the dominant frequency characteristics of the earthquake sources in each frequency band;

[0094] It can be understood that this step extracts the dominant frequency characteristics of the earthquake source in each frequency band by clustering analysis of the phase synchronization characteristics of the earthquake source. First, based on the previously obtained phase synchronization characteristics, the similarity measure between the frequency bands is determined. This measure can usually be based on phase consistency indicators, synchronization strength or similarity distance, etc. Then, the K-means clustering algorithm is used to group these frequency bands to identify frequency band groups with similar synchronization behaviors. After clustering, the frequency of each cluster center can be regarded as the dominant frequency characteristics of the corresponding frequency band, which represent the frequency components that best describe the characteristics of the earthquake source in the corresponding frequency band. In this way, cluster analysis helps to simplify the spectrum information of the earthquake source and extract the most representative frequency characteristics of the earthquake source.

[0095] Step S64: Based on the correlation analysis between the dominant frequency characteristics of the earthquake source and the historical response information of the building structure, by aligning the dominant frequency characteristics with the historical response information of the building structure, the correlation between the earthquake source frequency characteristics and the structural response is analyzed, and a building structure response prediction model is constructed.

[0096] It can be understood that this step uses a time series analysis method to align the dominant frequency characteristics of the earthquake source with the corresponding frequency bands in the historical structural response data, and analyze the amplitude changes and delayed responses of the signals in these frequency bands. The contribution of the earthquake source frequency to the response of the building structure can be quantitatively evaluated by correlation analysis, regression analysis or spectral analysis to reveal the response characteristics of the structure at a specific dominant frequency. Finally, combined with these data, a response prediction model is constructed using a support vector machine using statistical modeling or machine learning methods, so that it can predict the possible response of the building based on the input dominant frequency. In terms of technical effect, this model directly links the earthquake frequency characteristics with the structural response, helping to achieve early prediction of the building response under the influence of seismic waves. Among them, step S64 includes step S641, step S642 and step S643.

[0097] Step S641, performing Fourier transform on the historical response information of the building structure to obtain frequency domain information corresponding to the historical response information of the building structure;

[0098] It can be understood that in this step, the Fourier transform can decompose the complex time domain signals in the historical response data into a superposition of multiple sine waves, each of which corresponds to a response component of a specific frequency. In this way, we can identify the main response frequency of the building structure under historical earthquake events or environmental vibrations, as well as the contribution of each frequency component to the overall response. This frequency domain information is of great reference value for analyzing the response of the building structure under the dominant frequency of the earthquake source in the subsequent steps. By obtaining the frequency domain information of the historical response, the response characteristics of the structure at a specific frequency can be analyzed more intuitively, thus laying the foundation for building an accurate structural response prediction model.

[0099] Step S642: One-to-one correspondence is made between the frequency domain information corresponding to the historical response information of the building structure, and the frequency domain information corresponding to the historical response information of the building structure and the dominant frequency characteristics of the earthquake source are calculated based on the Pearson correlation coefficient to perform correlation analysis to obtain the correlation relationship between the frequency domain information corresponding to the historical response information of the building structure and the dominant frequency characteristics of the earthquake source;

[0100] It can be understood that in this step, the Pearson correlation coefficient can quantify the linear relationship between the response intensity of the building structure at different frequency components and the dominant frequency characteristics of the earthquake source. By calculating the correlation for each frequency component, we can observe whether the building structure has an obvious response at this frequency. This calculation can effectively identify the sensitivity of the building structure to the frequency of the seismic wave.

[0101] Step S643: The frequency domain information corresponding to the historical response information of the building structure and the dominant frequency characteristics of the earthquake source are input into a preset support vector regression model for training, wherein the association relationship is learned by the preset support vector regression model to obtain a trained building structure response prediction model.

[0102] It can be understood that in this step, the core idea of ​​the support vector regression model is to map the original data to a high-dimensional space and find a hyperplane so that the prediction error is as small as possible within a given tolerance range. The support vector regression model introduces a "tolerance" parameter to allow the prediction error of some data points to exceed a certain range, so that it can adapt to the noise in the data and prevent overfitting. The prediction method based on support vector regression has strong generalization ability and robustness, and can provide accurate prediction of building structure response under the background of different earthquake events, thereby providing strong support for earthquake disaster assessment and protection.

[0103] Step S7: predicting the building structure response in real time based on the building structure response prediction model, and issuing a real-time warning based on the prediction result.

[0104] It can be understood that in this step, first, the latest seismic wave image data is obtained through the real-time monitoring system, and the relevant source characteristics are extracted. These real-time data will be used as input to the trained building structure response prediction model. The model will combine historical response information and source characteristics to predict the possible response of the building structure under the current source conditions.

[0105] Next, the model calculates the real-time response indicators of the building structure based on the earthquake source and the dynamic characteristics of the building structure, which may include displacement, acceleration, velocity and other response quantities. The changes in these indicators can reflect whether the building structure has been subjected to stress or vibration beyond its safe bearing range.

[0106] Based on the predicted building structure response, the system will compare it with the preset safety threshold to determine whether the current building structure response exceeds the safety limit. If the response value exceeds the safety range, the system will trigger the early warning mechanism to notify relevant personnel to take protective measures or conduct emergency treatment.

[0107] Embodiment 2:

[0108] like Figure 2 As shown, this embodiment provides an analysis system based on seismic wave images, see Figure 2 The system includes an acquisition unit 701 , a processing unit 702 , an analysis unit 703 , a training unit 704 , a calculation unit 705 , an extraction unit 706 and an early warning unit 707 .

[0109] An acquisition unit 701 is used to acquire seismic wave image data, wherein the seismic wave image data includes acceleration data, displacement data, velocity data, timestamp data and direction data of the seismic wave;

[0110] A processing unit 702 is used to send the seismic wave image data to a preset feature extraction model for processing to obtain spatiotemporal feature data of the seismic wave image data;

[0111] An analysis unit 703 is used to analyze the spatiotemporal characteristic data of the seismic wave image data based on a preset multi-scale adaptive bandpass filter to obtain the initial estimated position information of the earthquake source corresponding to the seismic wave image data;

[0112] A training unit 704 is used to determine the historical response information of the building structure corresponding to the initial estimated position information of the earthquake source based on the initial estimated position information of the earthquake source, and send the initial estimated position information of the earthquake source and the historical response information of the building structure corresponding to the earthquake source to a preset deep autoencoder network for training to obtain a nonlinear response model;

[0113] A calculation unit 705 is used to calculate the fractal dimension of the seismic wave image data based on the nonlinear response model and a preset multi-fractal analysis method to obtain the influence intensity of the earthquake source at different spatial positions, and generate a spatial influence diagram of the earthquake source based on the influence intensity of the earthquake source at different spatial positions;

[0114] An extraction unit 706 is used to extract the dominant frequency characteristics of the earthquake source based on the spatial influence diagram of the earthquake source, and to construct a building structure response prediction model based on the extracted dominant frequency characteristics of the earthquake source and the historical response information of the building structure;

[0115] The early warning unit 707 is used to predict the building structure response in real time based on the building structure response prediction model, and to issue a real-time early warning based on the prediction result.

[0116] It should be noted that, regarding the system in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0117] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0118] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. An analysis method based on seismic wave images, characterized in that: include: Acquire seismic wave image data, wherein the seismic wave image data includes acceleration data, displacement data, velocity data, timestamp data and direction data of the seismic wave; Sending the seismic wave image data to a preset feature extraction model for processing to obtain spatiotemporal feature data of the seismic wave image data; Analyzing the spatiotemporal characteristic data of the seismic wave image data based on a preset multi-scale adaptive bandpass filter to obtain initial estimated position information of the earthquake source corresponding to the seismic wave image data; Determine the historical response information of the corresponding building structure based on the initial estimated position information of the earthquake source, and send the initial estimated position information of the earthquake source and the historical response information of the corresponding building structure to a preset deep autoencoder network for training to obtain a nonlinear response model; Based on the nonlinear response model and a preset multi-fractal analysis method, the fractal dimension of the seismic wave image data is calculated to obtain the influence intensity of the earthquake source at different spatial positions, and a spatial influence diagram of the earthquake source is generated based on the influence intensity of the earthquake source at different spatial positions; Extracting the dominant frequency characteristics of the earthquake source based on the spatial influence diagram of the earthquake source, and constructing a building structure response prediction model based on the extracted dominant frequency characteristics of the earthquake source and the historical response information of the building structure; The building structure response is predicted in real time based on the building structure response prediction model, and a real-time warning is given based on the prediction result.

2. The analysis method based on seismic wave images according to claim 1 is characterized in that , sending the seismic wave image data to a preset feature extraction model for processing to obtain spatiotemporal feature data of the seismic wave image data, including: Modeling the seismic wave image data based on a preset manifold learning method, wherein the domain relationship of each data point in the seismic wave image data and the distance between the data points are defined, and a manifold model is constructed based on the selected neighborhood relationship and the gradient information of each data point on the manifold model is calculated; Based on a preset t-SNE algorithm, the gradient information of each data point on the manifold model is mapped to a low-dimensional space for dimensionality reduction processing to obtain a spatiotemporal feature map of the seismic wave image data; Based on the principal component analysis method, the spatiotemporal feature data are extracted and processed from the spatiotemporal feature graph of the seismic wave image data to obtain the spatiotemporal feature data.

3. The analysis method based on seismic wave images according to claim 1 is characterized in that The method of analyzing the spatiotemporal characteristic data of the seismic wave image data based on a preset multi-scale adaptive bandpass filter includes: Constructing a bandpass filter based on the spatiotemporal feature data, wherein the center frequency and bandwidth of each bandpass filter are set by a window function to obtain at least two bandpass filters of different frequency bands; The above-mentioned spatiotemporal feature data is filtered based on the bandpass filters of different frequency bands, wherein all the bandpass filters are applied to the image data in the spatiotemporal feature data through convolution operation to obtain the image data of each frequency band after bandpass filtering; Calculating local maximum values ​​based on the image data of each frequency band after bandpass filtering, and comparing the local maximum values ​​with a preset threshold to obtain a comparison result; At least two candidate earthquake source positions are determined based on the comparison result, and all the candidate earthquake source positions are weightedly calculated to obtain an initial estimated earthquake source position corresponding to the seismic wave image data.

4. The analysis method based on seismic wave images according to claim 1 is characterized in that , based on the initial estimated position information of the earthquake source, the historical response information of the corresponding building structure is determined, and the initial estimated position information of the earthquake source and the historical response information of the corresponding building structure are sent to a preset deep autoencoder network for training to obtain a nonlinear response model, including: Based on the above-mentioned initial estimated position information of the earthquake source, a propagation path from the initial estimated position of the earthquake source to the building structure at the corresponding position is determined, and the propagation path of the initial estimated position of the earthquake source is compared with the propagation path in the historical records through a similarity matching algorithm to obtain the historical response information of the building structure corresponding to the initial estimated position of the earthquake source; The time-frequency characteristics of seismic wave image data are extracted based on segmented Fourier transform, and the historical response information of the building structure corresponding to the initial estimated location of the earthquake source is decomposed based on wavelet transform to obtain structural response characteristic data; The time-frequency feature and the structural response feature data are sent to a preset autoencoder network for training, wherein a mean square error function is used as a loss function, and the model parameters are updated by a stochastic gradient descent algorithm to obtain a trained nonlinear response model.

5. The analysis method based on seismic wave images according to claim 1 is characterized in that , based on the nonlinear response model and the preset multi-fractal analysis method, the fractal dimension of the seismic wave image data is calculated to obtain the influence intensity of the earthquake source at different spatial positions, and the spatial influence diagram of the earthquake source is generated based on the influence intensity of the earthquake source at different spatial positions, including: The seismic wave image data is evenly divided according to the initial estimated position of the earthquake source to obtain at least two seismic wave sub-images; Calculate the entropy change rate of all seismic wave sub-images to obtain the entropy change rate corresponding to each seismic wave sub-image; Determine structural response characteristic data of each seismic wave sub-image based on the nonlinear response model, and then obtain intensity distribution data of different sub-regions; The fractal dimension of the intensity distribution data of all sub-regions is calculated to obtain the local fractal dimension of each sub-region; Spatial interpolation is performed based on the local fractal dimensions of all sub-regions, and the fractal dimension of each sub-region obtained by interpolation is normalized to generate the spatial influence map of the earthquake source.

6. An analysis system based on seismic wave images, characterized in that: include: An acquisition unit, used for acquiring seismic wave image data, wherein the seismic wave image data includes acceleration data, displacement data, velocity data, time stamp data and direction data of the seismic wave; A processing unit, used for sending the seismic wave image data to a preset feature extraction model for processing to obtain spatiotemporal feature data of the seismic wave image data; An analysis unit, configured to analyze the spatiotemporal characteristic data of the seismic wave image data based on a preset multi-scale adaptive bandpass filter to obtain initial estimated position information of a seismic source corresponding to the seismic wave image data; A training unit, used to determine the historical response information of the corresponding building structure based on the initial estimated position information of the earthquake source, and send the initial estimated position information of the earthquake source and the historical response information of the corresponding building structure to a preset deep autoencoder network for training to obtain a nonlinear response model; A calculation unit, configured to calculate the fractal dimension of the seismic wave image data based on the nonlinear response model and a preset multi-fractal analysis method, obtain the influence intensity of the seismic source at different spatial positions, and generate a spatial influence diagram of the seismic source based on the influence intensity of the seismic source at different spatial positions; An extraction unit, used for extracting the dominant frequency characteristics of the earthquake source based on the spatial influence diagram of the earthquake source, and constructing a building structure response prediction model based on the extracted dominant frequency characteristics of the earthquake source and the historical response information of the building structure; The early warning unit is used to predict the building structure response in real time based on the building structure response prediction model, and to issue a real-time early warning based on the prediction result.

7. The seismic wave image-based analysis system according to claim 6, characterized in that: The processing unit comprises: A first processing subunit is used to model the seismic wave image data based on a preset manifold learning method, wherein the neighborhood relationship of each data point in the seismic wave image data and the distance between the data points are defined, and a manifold model is constructed based on the selected neighborhood relationship and the gradient information of each data point on the manifold model is calculated; A second processing subunit is used to map the gradient information of each data point on the manifold model to a low-dimensional space for dimensionality reduction based on a preset t-SNE algorithm to obtain a spatiotemporal feature map of the seismic wave image data; The third processing subunit is used to perform spatiotemporal feature data extraction processing on the spatiotemporal feature graph of the seismic wave image data based on the principal component analysis method to obtain spatiotemporal feature data.

8. The seismic wave image-based analysis system according to claim 6, characterized in that: The analysis unit comprises: A first analysis subunit is used to construct a bandpass filter based on the spatiotemporal feature data, wherein the center frequency and bandwidth of each bandpass filter are set by a window function to obtain at least two bandpass filters of different frequency bands; A second analysis subunit is used to filter the spatiotemporal feature data based on the bandpass filters of different frequency bands, wherein all the bandpass filters are applied to the image data in the spatiotemporal feature data through a convolution operation to obtain the image data of each frequency band after the bandpass filtering; A third analysis subunit is used to calculate the local maximum value based on the image data of each frequency band after the bandpass filtering, and compare the local maximum value with a preset threshold value to obtain a comparison result; The fourth analysis subunit is used to determine at least two candidate earthquake source locations based on the comparison result, and perform weighted calculation on all the candidate earthquake source locations to obtain an initial estimated earthquake source location corresponding to the seismic wave image data.

9. The seismic wave image-based analysis system according to claim 6, characterized in that: The training unit comprises: The first training subunit is used to determine the propagation path from the initial estimated position of the earthquake source to the building structure at the corresponding position based on the initial estimated position information of the earthquake source, and compare the propagation path of the initial estimated position of the earthquake source with the propagation path in the historical records through a similarity matching algorithm to obtain the historical response information of the building structure corresponding to the initial estimated position of the earthquake source; The second training subunit is used to extract the time-frequency characteristics of the seismic wave image data based on the segmented Fourier transform, and decompose the historical response information of the building structure corresponding to the initial estimated position of the earthquake source based on the wavelet transform to obtain the structural response characteristic data; The third training subunit is used to send the time-frequency characteristics and the structural response characteristic data to a preset autoencoder network for training, wherein a mean square error function is used as a loss function, and the model parameters are updated by a stochastic gradient descent algorithm to obtain a trained nonlinear response model.

10. The seismic wave image-based analysis system according to claim 6, characterized in that: The computing unit comprises: A first calculation subunit is used to evenly divide the seismic wave image data according to the initial estimated location of the earthquake source to obtain at least two seismic wave sub-images; The second calculation subunit is used to calculate the entropy change rate of all seismic wave sub-images to obtain the entropy change rate corresponding to each seismic wave sub-image; A third calculation subunit is used to determine the structural response characteristic data of each seismic wave sub-image based on the nonlinear response model, and then obtain the intensity distribution data of different sub-regions; A fourth calculation subunit is used to calculate the fractal dimension of the intensity distribution data of all sub-regions to obtain the local fractal dimension of each sub-region; The fifth calculation subunit is used to perform spatial interpolation based on the local fractal dimensions of all sub-regions, and normalize the fractal dimension of each sub-region obtained by interpolation, so as to generate a spatial influence map of the earthquake source.

Images