A method for classifying vibration waves, an electronic device, and a storage medium
Through the feature fusion method of short-time Fourier transform and multimodal neural network, the problem of poor adaptability of vibration wave classification is solved, and more accurate automatic vibration wave recognition is achieved.
Patent Information
- Application Number
- CN202510398950.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-01
- Publication Date
- 2025-07-04
- Estimated Expiration
- 2045-04-01
AI Technical Summary
In the prior art, the vibration wave classification method relies on waveforms of a single nature, resulting in poor adaptability, difficulty in achieving automated identification, and insufficient classification accuracy.
Short-time Fourier transform STFT is used to convert vibration waves into frequency domain feature matrix, combining characteristic parameters such as maximum vibration velocity and average vibration velocity, and feature fusion is performed through multimodal neural networks, and the decision layer is used to output vibration wave categories.
It enriches the characteristic information of vibration waves, improves the accuracy and generality of classification results, and adapts to vibration wave classification in different regions.
Smart Images
Figure CN119903399B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of seismic wave classification, and in particular, to a seismic wave classification method, an electronic device, and a storage medium. Background Art
[0002] Seismic wave classification is of great significance for seismic hazard assessment, regional tectonic research, and the seismogenic mechanism of induced earthquakes. In traditional seismic monitoring, technicians identify the types of seismic events through parameters such as P-wave polarity, spectral characteristics, and moment tensors. This identification method relies on manual experience and is difficult to automate. To solve the above problems, in the prior art, feature extraction is performed on seismic waves, and then the extracted features are input into a trained model for automatic identification of the types of seismic waves. Since this method is based on waveforms of a single nature, its adaptability is poor, and there are significant differences in the accuracy of classifying seismic waves in different regions, resulting in low accuracy of the classification results of seismic waves. Summary of the Invention
[0003] In view of the above technical problems, the technical solution adopted by the present invention is as follows:
[0004] According to a first aspect of the present application, a seismic wave classification method is provided. The method includes the following steps:
[0005] S100, obtaining an original seismic wave; wherein, the original seismic wave is a time series signal;
[0006] S200, using the short-time Fourier transform (STFT) to convert the original seismic wave into a corresponding frequency-domain feature matrix; wherein, the frequency-domain feature matrix includes a plurality of rows and a plurality of columns, each row corresponds to a frequency, and each column corresponds to a time period; the element value of the frequency-domain feature matrix is the modulus of the amplitude of the seismic wave corresponding to the corresponding time period and the corresponding frequency;
[0007] S300, obtaining characteristic parameters corresponding to the original seismic wave in each preset direction; wherein, the characteristic parameters include the maximum vibration velocity and the average vibration velocity of the original seismic wave;
[0008] S400, inputting the frequency-domain feature matrix corresponding to the original seismic wave into a first encoder of a preset seismic wave classification model to obtain an n-dimensional first feature vector corresponding to the frequency-domain feature matrix;
[0009] S500, inputting the original seismic wave into a second encoder of the preset seismic wave classification model to obtain an n-dimensional second feature vector corresponding to the original seismic wave;
[0010] S600, inputting the characteristic parameters corresponding to the original seismic wave in each preset direction into a third encoder of the preset seismic wave classification model to obtain an n-dimensional third feature vector corresponding to the original seismic wave;
[0011] S700, perform feature fusion on the first eigenvector, the second eigenvector, and the third eigenvector to obtain a fused eigenvector;
[0012] S800, input the fused eigenvector into the decision layer to obtain the category corresponding to the original vibration wave.
[0013] According to another aspect of the present application, there is also provided a non-transitory computer-readable storage medium storing at least one instruction or at least one program segment, and the at least one instruction or at least one program segment is loaded and executed by a processor to implement the above vibration wave classification method.
[0014] According to another aspect of the present application, there is also provided an electronic device including a processor and the above non-transitory computer-readable storage medium.
[0015] The present invention has at least the following beneficial effects:
[0016] In the vibration wave classification method of the present invention, the short-time Fourier transform (STFT) is used to convert the original vibration wave into a frequency domain feature matrix corresponding to the original vibration wave; the characteristic parameters corresponding to the original vibration wave in each preset direction are obtained; the frequency domain feature matrix corresponding to the original vibration wave is input into the first encoder of a preset vibration wave classification model to obtain an n-dimensional first eigenvector corresponding to the frequency domain feature matrix; the original vibration wave is input into the second encoder of the preset vibration wave classification model to obtain an n-dimensional second eigenvector corresponding to the original vibration wave; the characteristic parameters corresponding to the original vibration wave in each preset direction are input into the third encoder of the preset vibration wave classification model to obtain an n-dimensional third eigenvector corresponding to the original vibration wave; perform feature fusion on the first eigenvector, the second eigenvector, and the third eigenvector to obtain a fused eigenvector, and then output the category corresponding to the original vibration wave through the decision layer; during the classification process of the vibration wave, since the time domain features, frequency domain features, and characteristic parameters of the original vibration wave are respectively obtained and the frequency domain features, time domain features, and characteristic parameters of the original vibration wave are fused, the features of the original vibration wave are made more abundant, and the fused eigenvector after fusion can more accurately reflect the characteristics of the original vibration wave. Therefore, the final classification result is more accurate. Description of the Drawings
[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings without creative efforts based on these drawings.
[0018] Figure 1Flowchart of the vibration wave classification method provided by an embodiment of the present invention;
[0019] Figure 2 Structural schematic diagram of the preset vibration wave classification model provided by an embodiment of the present invention. Detailed implementation manners
[0020] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts fall within the protection scope of the present invention.
[0021] It should be noted that based on this disclosure, those skilled in the art should understand that one aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number of aspects described herein can be used to implement the device and / or practice the method. In addition, this device can be implemented and this method can be practiced using other structures and / or functions in addition to one or more of the aspects described herein.
[0022] Next, a vibration wave classification method will be introduced with reference to Figure 1 the flowchart of the vibration wave classification method shown.
[0023] The vibration wave classification method may include the following steps:
[0024] S100, obtaining an original vibration wave; wherein, the original vibration wave is a time series signal.
[0025] In this embodiment, when a natural earthquake, a mine earthquake or a collapse occurs, an original vibration wave can be obtained; it can be understood that the original vibration wave is not necessarily generated by a natural earthquake, and it may also be generated by a mine earthquake or a collapse; usually, the obtained original vibration wave is a time series signal.
[0026] S200, using the short-time Fourier transform (STFT) to convert the original vibration wave into a corresponding frequency domain feature matrix; wherein, the frequency domain feature matrix includes several rows and several columns, each row corresponds to a frequency, and each column corresponds to a time period; the element value of the frequency domain feature matrix is the modulus of the vibration wave amplitude corresponding to the time period and the corresponding frequency.
[0027] In this embodiment, the short-time Fourier transform can be used to process the original vibration wave, dividing the original vibration wave into multiple time periods and performing Fourier transform on each time period, so as to obtain the spectral characteristics of the original vibration wave changing with time; taking the modulus of the vibration wave amplitude at each frequency as an element value, so that the modulus of the vibration wave amplitude at different frequencies corresponding to each time period can be obtained, and then the frequency domain feature matrix can be obtained.
[0028] S300. Obtain the characteristic parameters corresponding to the original vibration wave in each preset direction; wherein, the characteristic parameters include the maximum vibration speed and the average vibration speed of the original vibration wave.
[0029] In this embodiment, the preset directions include the vertically downward direction, the north-south direction, and the east-west direction; the characteristic parameters include the maximum vibration speed P V and the average vibration speed IAV; the calculation formulas for the characteristic parameters are as follows:
[0030] ;
[0031] ;
[0032] wherein, v(t) is the time function of the original vibration wave; pt is the preset time point.
[0033] For example: The original vibration wave is the waveform data of 120 s, and the input dimension is (3, 12000); the frequency domain feature matrix reflects the change characteristics in the frequency domain, and the input dimension is (3, 200, 1201); the characteristic parameters corresponding to the original vibration wave in each preset direction provide additional background information for the model, and the dimension is (1, 6).
[0034] Through the above formula, the maximum vibration speeds P v，UD , P v，NS and P v，EW in the vertically downward direction, the north-south direction, and the east-west direction of the original vibration wave can be obtained; and the average vibration speeds IAV UD , IAV NS and IAV EW in the vertically downward direction, the north-south direction, and the east-west direction of the original vibration wave.
[0035] S400. Input the frequency domain feature matrix corresponding to the original vibration wave into the first encoder of the preset vibration wave classification model to obtain the n-dimensional first feature vector corresponding to the frequency domain feature matrix.
[0036] In this embodiment, in order to process the two sets of features in the time domain and the frequency domain, a multi-modal neural network can be constructed to process these two types of features respectively:
[0037] Further, as Figure 2As shown in the figure, the first encoder of the preset vibration wave classification model includes: a number of one-dimensional CNN convolutional layers and an encoding layer; among them, a number of one-dimensional CNN convolutional layers Conv1D are used to extract features from the frequency domain feature matrix, and the encoding layer is used to encode the features of the extracted frequency domain feature matrix.
[0038] S500, input the original vibration wave into the second encoder of the preset vibration wave classification model to obtain an n-dimensional second feature vector corresponding to the original vibration wave.
[0039] In this embodiment, the second encoder of the preset vibration wave classification model includes: a number of two-dimensional convolutional layers Conv2D; among them, a number of two-dimensional convolutional layers are used to extract time domain and frequency domain features from the original vibration wave.
[0040] S600, input the feature parameters corresponding to the original vibration wave in each preset direction into the third encoder of the preset vibration wave classification model to obtain an n-dimensional third feature vector corresponding to the original vibration wave.
[0041] In this embodiment, the third encoder of the preset vibration wave classification model includes: a number of linear layers FC.
[0042] The original vibration wave input, Waveform input, first extracts information through one-dimensional CNN convolution, and then performs feature encoding through the Transformer encoding layer. The Transformer structure is widely used in seismology due to its information processing ability for time series signals. The input of the frequency domain feature matrix, that is, the input of the spectrogram STFT input, extracts time-frequency domain features through two-dimensional convolution. The ReLU activation function is connected between the convolutional layers to introduce non-linearity and improve the calculation efficiency and network expression ability. The input of the feature parameters, Feature input, extracts features through the linear layer FC.
[0043] S700, perform feature fusion on the first feature vector, the second feature vector and the third feature vector to obtain a fused feature vector.
[0044] Further, step S700 includes the following steps:
[0045] S710, obtain the first feature vector XA = (XA1, XA2,..., XA i ,..., XA n ), i = 1, 2,..., n; where XA i is the i-th frequency domain feature corresponding to the frequency domain feature matrix.
[0046] S711, obtain the second feature vector XB = (XB1, XB2,..., XB i ,..., XB n ); where XBi The i-th time-frequency domain feature corresponding to the original vibration wave.
[0047] S712, obtain the third feature vector XC = (XC1, XC2,..., XC i ,..., XC n ); where XC i is the i-th feature parameter corresponding to the original vibration wave.
[0048] S713, splice XA, XB, and XC to obtain a fused feature vector XD = (XA1, XA2,..., XA i ,..., XA n , XB1, XB2,..., XB i ,..., XB n , XC1, XC2,..., XC i ,..., XC n ).
[0049] In this embodiment, after obtaining the first feature vector corresponding to the frequency domain feature matrix, the second feature vector corresponding to the original vibration wave, and the third feature vector, the first feature vector, the second feature vector, and the third feature vector can be quickly fused by splicing to obtain a fused feature vector. The fused feature vector includes the frequency domain feature, the time domain feature, and the feature parameters of the original vibration wave, and can better reflect the characteristics of the original vibration wave.
[0050] Further, after step S712, the method further includes the following steps:
[0051] S720, obtain each preset geographical feature of the target area where the original vibration wave is generated to obtain a geographical feature list D = (D1, D2,..., D j ,..., D m ), j = 1, 2,..., m; where D j is the j-th preset geographical feature of the target area where the original vibration wave is generated, and m is the number of preset geographical features; the preset geographical features include altitude, terrain type, landform, underground rock layer thickness, and soil type.
[0052] In this embodiment, it can be understood that the generation of the original vibration wave corresponds to an area. However, the geographical features of different areas are also different. For example, there are significant differences in altitude, terrain type, landform, underground rock layer thickness, and soil type between the first area and the second area that are far apart; it should be noted that the characteristics of the vibration waves generated by the same event in different areas may also be significantly different.
[0053] S730. Convert each geographical feature in D into a corresponding feature value to obtain the geographical feature vector D' = (D'1, D'2,..., D' j ,..., D' m ); where D' j is the corresponding feature value of D j .
[0054] Furthermore, D' j can be obtained through the following steps:
[0055] S731. If D j is a preset numerical feature, then determine D' j = D j .
[0056] In this embodiment, the numerical feature can be understood as a feature with a specific numerical parameter, such as altitude, underground rock layer thickness, etc.
[0057] S732. If D j is a preset type feature, then obtain the corresponding feature value mapping table YD j of D j ; where YD j includes several rows, and each row corresponds to a type feature and the feature value corresponding to the type feature.
[0058] S733. According to YD j , determine the corresponding feature value D' j of D j .
[0059] In this embodiment, the type feature can be understood as a feature whose parameter is not a specific numerical value, such as the terrain type is plain, the soil type is clay type, etc.; all types under each feature can be listed in an exhaustive manner, and then each type is uniquely encoded to obtain the corresponding feature value mapping table; when used later, only need to traverse the feature mapping table to determine the corresponding feature value.
[0060] S740. Perform feature fusion on the first feature vector, the second feature vector, and the third feature vector according to D'.
[0061] Furthermore, step S740 can include the following steps:
[0062] S741. Input D' into a preset fusion weight prediction model to obtain the fusion weight list W = (W1, W2,..., W k ,..., W 3n ), k = 1, 2,..., 3n; where W kis the k-th weight obtained; 3n is the sum of elements in the first eigenvector, the second eigenvector, and the third eigenvector.
[0063] In this embodiment, a fusion weight prediction model is preset, and the fusion weight prediction model is used to output a list of fusion weights corresponding to different geographical features.
[0064] S742, according to W, perform feature fusion on the first eigenvector, the second eigenvector, and the third eigenvector to obtain a fused eigenvector XD = (W1×XA1, W2×XA2,..., W i ×XA i ,..., W n ×XA n , W n+1 ×XB1, W n+2 ×XB2,..., W n+i ×XB i ,..., W 2n ×XB n , W 2n+1 ×XC1, W 2n+2 ×XC2,..., W 2n+i ×XC i ,..., W 3n ×XC n ).
[0065] In this embodiment, through the above method, the geographical features of the region where the original vibration wave is generated can be added to the first feature and the second feature, so that the obtained fused feature not only contains the frequency domain feature and the time domain feature of the original vibration wave, but also contains the geographical features of the region where the original vibration wave is generated, thereby improving the versatility of the vibration wave classification model and making the final vibration wave classification result more accurate.
[0066] Further, the output layer of the preset vibration wave classification model includes Dense and Softmax.
[0067] In this embodiment, for example: there are 3 types of original vibration waves: natural earthquakes, mine tremors, and collapses. Then Dense outputs three results, which are normalized by Softmax to obtain the final classification result.
[0068] Further, n = 64.
[0069] In this embodiment, the Adam optimizer and the schedule mechanism are used for model training to ensure that the learning rate can be adjusted efficiently during the optimization process. The Adam optimizer, with its adaptive learning rate and momentum update mechanism, can dynamically adjust the learning rate when updating different parameters, thereby accelerating convergence and improving model performance. We set the maximum number of training epochs to 100 and adopt an early stopping mechanism with patience set to 10 to implement the early stopping strategy and prevent overfitting. That is, if there is no improvement in the validation loss for 10 consecutive epochs, the training will stop prematurely.
[0070] During the training process, the multi-class cross-entropy loss function is used, which can effectively measure the difference between the model's predicted probability and the true label. By monitoring the loss value and learning curve during the training process, the convergence situation and performance changes of the model can be clearly observed. The training learning curve shows the change trend of the loss function in each training epoch, thus achieving a better training effect.
[0071] The vibration wave classification method of this embodiment uses the short-time Fourier transform (STFT) to convert the original vibration wave into a frequency-domain feature matrix corresponding to the original vibration wave; obtains the characteristic parameters corresponding to the original vibration wave in each preset direction; inputs the frequency-domain feature matrix corresponding to the original vibration wave into the first encoder of the preset vibration wave classification model to obtain an n-dimensional first feature vector corresponding to the frequency-domain feature matrix; inputs the original vibration wave into the second encoder of the preset vibration wave classification model to obtain an n-dimensional second feature vector corresponding to the original vibration wave; inputs the characteristic parameters corresponding to the original vibration wave in each preset direction into the third encoder of the preset vibration wave classification model to obtain an n-dimensional third feature vector corresponding to the original vibration wave; performs feature fusion on the first feature vector, the second feature vector, and the third feature vector to obtain a fused feature vector, and then outputs the category corresponding to the original vibration wave through the decision layer; during the classification process of the vibration wave, since the time-domain features, frequency-domain features, and characteristic parameters of the original vibration wave are respectively obtained and the frequency-domain features, time-domain features, and characteristic parameters of the original vibration wave are fused, the features of the original vibration wave become richer, and the fused feature vector after fusion can more accurately reflect the characteristics of the original vibration wave. Therefore, the final classification result is more accurate.
[0072] In addition, although the steps of the methods in the present disclosure are described in a specific order in the drawings, this does not require or imply that these steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step for execution, and / or one step may be decomposed into multiple steps for execution, etc.
[0073] Embodiments of the present invention also provide a non-transitory computer-readable storage medium, which can be disposed in an electronic device to store at least one instruction or at least one segment of a program related to a method in a method embodiment. The at least one instruction or the at least one segment of the program is loaded and executed by the processor to implement the method provided in the above embodiments.
[0074] The program product can adopt any combination of one or more readable media. The readable media can be a readable signal medium or a readable storage medium. The readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of the readable storage medium include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.
[0075] The computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, which carries the readable program code. Such a propagated data signal can take various forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The readable signal medium can also be any readable medium other than the readable storage medium, which can send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.
[0076] The program code contained on the readable medium can be transmitted by any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination of the above.
[0077] The program code for performing the operations of the present application can be written in any combination of one or more programming languages. The programming languages include object-oriented programming languages such as Java, C++, etc., and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computing device, partially on the user's device, executed as an independent software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user's computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, by using an Internet service provider to connect through the Internet).
[0078] Embodiments of the present invention also provide an electronic device, including a processor and the aforementioned non-transitory computer-readable storage medium.
[0079] The electronic device is merely an example and should not impose any limitations on the functions and usage scope of the embodiments of the present application.
[0080] The electronic device is presented in the form of a general-purpose computing device. The components of the electronic device may include, but are not limited to: at least one of the aforementioned processors, at least one of the aforementioned memories, and a bus connecting different system components (including the memory and the processor).
[0081] Among them, the memory stores program code, and the program code can be executed by the processor, so that the processor executes the steps in various embodiments described in this specification.
[0082] The memory may include a readable medium in the form of volatile memory, such as random access memory (RAM) and / or cache memory, and may further include read-only memory (ROM).
[0083] The memory may also include a program / utilities having a set (at least one) of program modules. Such program modules include, but are not limited to: an operating system, one or more application programs, other program modules, and program data. Implementations of a network environment may be included in each or some combination of these examples.
[0084] The bus may represent one or more of several types of bus structures, including a memory bus or a memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus using any of the various bus structures.
[0085] The electronic device may also communicate with one or more external devices (such as a keyboard, a pointing device, a Bluetooth device, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device, and / or may communicate with any device that enables the electronic device to communicate with one or more other computing devices (such as a router, a modem, etc.). Such communication may be carried out through an input / output (I / O) interface. Moreover, the electronic device may also communicate with one or more networks (such as a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) through a network adapter. The network adapter communicates with other modules of the electronic device through the bus. It should be understood that although not shown in the figure, other hardware and / or software modules may be used in combination with the electronic device, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.
[0086] Through the description of the above embodiments, those skilled in the art can easily understand that the exemplary embodiments described herein can be implemented by software or by a combination of software and necessary hardware. Therefore, the technical solution according to the embodiments of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (such as a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the embodiments of the present disclosure.
[0087] The embodiments of the present invention also provide a computer program product, which includes program code. When the program product runs on an electronic device, the program code is used to cause the electronic device to execute the steps in the methods according to various exemplary embodiments of the present invention described above in this specification.
[0088] Although some specific embodiments of the present invention have been described in detail by way of examples, those skilled in the art should understand that the above examples are only for the purpose of illustration and not for the purpose of limiting the scope of the present invention. Those skilled in the art should also understand that various modifications can be made to the embodiments without departing from the scope and spirit of the present invention.
Claims
1. A vibration wave classification method, characterized in that, The method includes the following steps: S100. Obtain an original vibration wave; wherein, the original vibration wave is a time series signal; S200. Use the short-time Fourier transform (STFT) to convert the original vibration wave into a corresponding frequency domain feature matrix; wherein, the frequency domain feature matrix includes a plurality of rows and a plurality of columns, each row corresponds to a frequency, and each column corresponds to a time period; the element value of the frequency domain feature matrix is the modulus of the vibration wave amplitude corresponding to the corresponding time period and the corresponding frequency; S300. Obtain the characteristic parameters corresponding to the original vibration wave in each preset direction; wherein, the characteristic parameters include the maximum vibration speed and the average vibration speed of the original vibration wave; S400. Input the frequency domain feature matrix corresponding to the original vibration wave into the first encoder of a preset vibration wave classification model to obtain an n-dimensional first feature vector corresponding to the frequency domain feature matrix; S500. Input the original vibration wave into the second encoder of the preset vibration wave classification model to obtain an n-dimensional second feature vector corresponding to the original vibration wave; S600. Input the characteristic parameters corresponding to the original vibration wave in each preset direction into the third encoder of the preset vibration wave classification model to obtain an n-dimensional third feature vector corresponding to the original vibration wave; S700. Perform feature fusion on the first feature vector, the second feature vector, and the third feature vector to obtain a fused feature vector; S800. Input the fused feature vector into a decision layer to obtain the category corresponding to the original vibration wave; Step S700 includes the following steps: S710. Obtain the first feature vector XA; S711. Obtain the second feature vector XB; S712. Obtain the third feature vector XC; S713. Concatenate XA, XB, and XC to obtain a fused feature vector XD; After step S712, the method further includes the following steps: S720. Obtain each preset geographical feature of the target area where the original vibration wave is generated to obtain a geographical feature list D corresponding to the original vibration wave; the preset geographical features include altitude, terrain type, landform, underground rock layer thickness, and soil type; S730. Convert each geographical feature in D into a corresponding feature value to generate a geographical feature vector D' of the target area of the original vibration wave; S740. Perform feature fusion on XA, XB, and XC according to D'; Step S740 includes the following steps: S741. Input D' into a preset fusion weight prediction model to obtain a fusion weight list W corresponding to XA, XB, and XC; S742. Perform feature fusion on XA, XB, and XC according to W to obtain a fused feature vector XD.
2. The shock wave classification method according to claim 1, wherein D’ j obtained through the following steps: S731, if D j is a preset numerical feature, then determine D' j = D j ; S732, if D j is a preset type feature, then obtain D j corresponding eigenvalue mapping table YD j ; among them, YD j includes several rows, each row corresponding to a type feature and the eigenvalue corresponding to the type feature; S733, determine D according to YD j , and determine the corresponding eigenvalue D' of D j j . 3. The shock wave classification method according to claim 1, wherein The first encoder of the preset vibration wave classification model includes: a plurality of one-dimensional CNN convolutional layers and an encoding layer; wherein, the plurality of one-dimensional CNN convolutional layers are used to extract features from the frequency domain feature matrix, and the encoding layer is used to perform feature encoding on the features of the extracted frequency domain feature matrix; The second encoder of the preset vibration wave classification model includes: a plurality of two-dimensional convolutional layers; wherein, the plurality of two-dimensional convolutional layers are used to extract time domain and frequency domain features from the original vibration wave; The third encoder of the preset vibration wave classification model includes: a plurality of linear layers.
4. The shock wave classification method according to claim 1, characterized in that n=64。 5. A non-transitory computer-readable storage medium storing at least one instruction or at least one program segment, characterized in that, The at least one instruction or the at least one program segment is loaded and executed by a processor to implement the shock wave classification method according to any one of claims 1-4.
6. An electronic device, characterized in that, Comprising a processor and the non-transitory computer-readable storage medium according to claim 5.
Citation Information
Patent Citations
Micro-seismic signal spectrum analysis system and method based on DVEC
CN119247464A