A background noise eicofunctional imaging method based on physical information neural network
Through the background noise Chenghan imaging method based on physical information neural network, the problems of high computer storage space consumption and velocity details loss in dense platform seismic recording are solved, and efficient and accurate underground velocity structure recovery is achieved.
Patent Information
- Application Number
- CN202211276137.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-10-18
- Publication Date
- 2025-08-26
- Estimated Expiration
- 2042-10-18
AI Technical Summary
The classic Cheng Han imaging method consumes a lot of computer storage space in dense platform seismic recording, and the linear interpolation algorithm loses the velocity details between station pairs, making it impossible to effectively utilize dense array data.
The background noise journey function imaging method based on physical information neural network is adopted. By acquiring the original seismic data for preprocessing, cross-correlation background noise to obtain the time-lapse data, a loss function containing physical constraints is constructed and the neural network is trained to directly obtain the underground velocity structure information.
It effectively avoids the iterative process of nonlinear inversion, reduces the amount of data storage, and uses the combination of process function equations and neural networks to accurately restore the speed details between stations, and the results are more reliable.
Smart Images
Figure CN115903018B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of seismic tomography, and in particular to a background noise eikonal imaging method based on a physical information neural network. Background Art
[0002] Seismic tomography is a method of understanding the Earth's internal structure by using physical signals (such as sound or electromagnetic waves) that penetrate an object. This method has the advantage of revealing its geometry or physical properties without damaging the object itself. In seismology, tomography can recover parameters such as velocity, anisotropy, and attenuation coefficient of the subsurface medium. It is a crucial tool for intuitively understanding subsurface structure and is widely used for imaging at various scales, including local, regional, and global.
[0003] Among them, seismic background noise Eikonal tomography has gradually become an important tool for noise imaging with the development of dense arrays. Background noise tomography based on the Eikonal equation treats each station as a virtual "seismic source", and the travel-time surface between this "seismic source" and other stations can be calculated. The Eikonal equation intuitively constructs the relationship between seismic wave velocity and propagation direction and spatial gradient, avoiding the conventional forward and inversion processes. However, classic Eikonal imaging uses a linear interpolation algorithm to reconstruct the travel-time surface between station pairs to estimate the velocity value in the Eikonal equation. This interpolation method often loses the velocity details between station pairs. In addition, classic Eikonal imaging requires the acquisition of cross-correlated travel-time information between all station pairs, which uses a large amount of data and consumes computer storage space when the amount of information recorded by dense array seismic records increases dramatically. Summary of the Invention
[0004] In order to solve the above problems, an embodiment of the present application provides a background noise einofunctional imaging method based on a physical information neural network.
[0005] In a first aspect, an embodiment of the present application provides a background noise eicofunctional imaging method based on a physical information neural network, the method comprising:
[0006] Acquiring original seismic data and performing a first preprocessing on the original seismic data;
[0007] Cross-correlating background noise in the raw seismic data to obtain travel time data between at least one set of station pairs;
[0008] Constructing an artificial neural network, constructing a loss function including physical constraints based on the eikonal equation, and adding the loss function to the artificial neural network;
[0009] After performing a second preprocessing on the training data, the artificial neural network is trained based on the training data to obtain an initial neural network model, wherein the training data includes the original seismic data and travel time data;
[0010] Target seismic data of a target area is acquired, and the target seismic data is processed based on the initial neural network model to obtain underground velocity structure information of the target area.
[0011] Preferably, the first preprocessing of the original seismic data includes:
[0012] The raw seismic data are sequentially subjected to format conversion, mean removal, instrument response removal, data resampling, and filtering.
[0013] Preferably, cross-correlating background noise in the original seismic data to obtain travel time data between at least one set of station pairs includes:
[0014] The original seismic data is divided based on frequency bands, and background noise in the original seismic data of each frequency band is cross-correlated to obtain travel time data between at least one set of station pairs.
[0015] Preferably, the second preprocessing of the training data includes:
[0016] Convert geographic coordinate system to Cartesian coordinate system based on Mercator projection;
[0017] Standardize and normalize the training data;
[0018] Coordinate system grid division is performed based on the station distance between each station and the data distribution of the input data.
[0019] Preferably, the initial neural network model includes a travel time neural network and a velocity neural network;
[0020] The step of acquiring target seismic data of a target area and processing the target seismic data based on the initial neural network model to obtain underground velocity structure information of the target area includes:
[0021] Acquire target seismic data of a target area, and process the target seismic data based on the travel time neural network and the velocity neural network to obtain target travel time data and target velocity data;
[0022] The underground velocity structure information of the target area is generated based on the target travel time data and the target velocity data.
[0023] In a second aspect, an embodiment of the present application provides a background noise eikonal imaging device based on a physical information neural network, the device comprising:
[0024] an acquisition module, configured to acquire original seismic data and perform a first preprocessing on the original seismic data;
[0025] A cross-correlation module, configured to cross-correlate background noise in the original seismic data to obtain travel time data between at least one set of station pairs;
[0026] A construction module, configured to construct an artificial neural network, construct a loss function including physical constraints based on the eikonal equation, and add the loss function to the artificial neural network;
[0027] A training module, configured to perform a second preprocessing on the training data and then train the artificial neural network based on the training data to obtain an initial neural network model, wherein the training data includes the original seismic data and travel time data;
[0028] The prediction module is used to obtain target seismic data of the target area, process the target seismic data based on the initial neural network model, and obtain underground velocity structure information of the target area.
[0029] In a third aspect, an embodiment of the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and runnable on the processor, wherein when the processor executes the computer program, the steps of the method provided in the first aspect or any possible implementation of the first aspect are implemented.
[0030] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method provided in the first aspect or any possible implementation of the first aspect.
[0031] The beneficial effects of the present invention are: 1. The use of the idea of physical information neural network structure combined with einofunctional imaging can effectively avoid the iterative process in nonlinear inversion, and the desired seismic wave velocity parameters can be directly obtained after one training.
[0032] 2. Only a small amount of travel time data is needed to train the network and predict the velocity. There is no need to treat all stations as earthquake sources for cross-correlation, which reduces the amount of data storage.
[0033] 3. In the process of inverting the velocity structure using the information contained in the traveltime data, the Eikonal equation and boundary conditions were added to impose physical constraints, making better use of the known information. In the process of restoring the traveltime surface, the neural network nonlinear system was used to more accurately obtain the velocity details between stations.
[0034] 4. The combination of the Eikonal equation and neural networks fully utilizes the known laws of seismic wave propagation, effectively preventing the neural network from learning false information caused by errors in the observation data, making the results more reliable. BRIEF DESCRIPTION OF THE DRAWINGS
[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0036] Figure 1 A schematic flow chart of a background noise eicofunctional imaging method based on a physical information neural network provided in an embodiment of the present application;
[0037] Figure 2 A schematic structural diagram of a background noise einofunctional imaging device based on a physical information neural network provided in an embodiment of the present application;
[0038] Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0039] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application.
[0040] In the following introduction, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance. The following introduction provides multiple embodiments of the present application. Different embodiments can be replaced or combined, so the present application can also be considered to include all possible combinations of the same and / or different embodiments described. Therefore, if one embodiment includes features A, B, and C, and another embodiment includes features B and D, then the present application should also be considered to include embodiments containing one or more of all other possible combinations of A, B, C, and D, even though the embodiment may not be clearly described in the following text.
[0041] The following description provides examples and does not limit the scope, applicability, or examples set forth in the claims. Changes may be made to the function and arrangement of the elements described without departing from the scope of the present application. Various examples may appropriately omit, replace, or add various processes or components. For example, the described method may be performed in an order different from the order described, and various steps may be added, omitted, or combined. In addition, features described in some examples may be combined in other examples.
[0042] See also Figure 1, Figure 1 : This is a flow chart of a method for background noise eicofunctional imaging based on a physical information neural network provided in an embodiment of the present application. In this embodiment of the present application, the method includes:
[0043] S101: Acquire original seismic data, and perform a first preprocessing on the original seismic data.
[0044] The execution entity of this application can be a cloud server.
[0045] In an embodiment of the present application, in order to construct and train a neural network model capable of predicting underground velocity structure, the cloud server will first obtain the original seismic data for training and preprocess the original seismic data before training to ensure the effectiveness of the training.
[0046] In one embodiment, the performing a first preprocessing on the raw seismic data includes:
[0047] The raw seismic data are sequentially subjected to format conversion, mean removal, instrument response removal, data resampling, and filtering.
[0048] In the embodiment of the present application, specifically, the raw seismic data will be format converted, de-meaned, de-instrumented, resampled, and filtered in sequence. Among them, format conversion is to convert the format level of data stored by different seismic instruments. There are many different storage methods for seismic data, such as SEED (Standard for the Exchange of Earthquake Data), miniSEED, SAC (Seismic Analysis Code), etc. The processing system needs to upgrade the original recording format to a high-level format for reading and processing; de-meaning is to remove the non-zero mean contained in the waveform information to avoid its impact on data analysis; de-instrumentation is to unify the differences in seismic records caused by different seismic instruments themselves and avoid signal errors caused by the instruments themselves. This can be achieved by matching the corresponding response function according to the RESP or PZ file of the instrument response; resampling is to obtain a suitable sampling period, unify the sampling period of different instruments or unequally spaced sampling data, and perform downsampling or interpolation according to research needs; filtering is to remove frequency information of non-target frequencies from the seismic record and limit the data to a certain frequency range for analysis.
[0049] S102: Cross-correlate background noise in the original seismic data to obtain travel time data between at least one set of station pairs.
[0050] In the embodiments of the present application, the raw seismic data contains background noise. By performing cross-correlation on this background noise, travel time data between pairs of stations can be calculated. Cross-correlation is the infinite integral of two functions, complex conjugated and inversely shifted, and then multiplied together. At the maximum value in the cross-correlation series, the two time series align. The index of this maximum value multiplied by the displacement time is the delay obtained by cross-correlation, which is the travel time data.
[0051] In one embodiment, cross-correlating background noise in the raw seismic data to obtain travel time data between at least one set of station pairs includes:
[0052] The original seismic data is divided based on frequency bands, and background noise in the original seismic data of each frequency band is cross-correlated to obtain travel time data between at least one set of station pairs.
[0053] In an embodiment of the present application, in order to further ensure the accuracy of subsequent prediction data results, the original seismic data can be divided according to frequency bands, and then the travel time data in different frequency bands can be calculated. In the subsequent process, the travel time data in different frequency bands can be inverted separately.
[0054] S103: construct an artificial neural network, construct a loss function containing physical constraints based on the eikonal equation, and add the loss function to the artificial neural network.
[0055] In the embodiments of the present application, neural networks, as nonlinear processing systems, are highly adaptable to geophysical nonlinear inversion processes. The fundamental concept of classic artificial neural networks (ANNs) is to mimic the way animal neurons process information, using a large number of interconnected neurons to form a complex computing system for distributed parallel information processing. They focus on converting raw data into information and then applying these characteristics or relationships to make predictions. Complex neural networks are capable of learning from the analysis of large datasets and automatically determining weight parameters that meet expectations. However, the design of classic neural networks often ignores the underlying physical principles driving inversion. These algorithms lack robustness, convergence, and interpretability with limited available training datasets. Physics-informed neural networks (PINNs) take into account the physical factors inherent in real data. They are neural networks trained to solve supervised learning tasks and can be used to describe any given physical law using general nonlinear partial differential equations. Using this network structure in conjunction with einofunctional imaging can effectively avoid the iterative process involved in nonlinear inversion and directly obtain the desired seismic velocity parameters with a single training run.
[0056] Specifically, an artificial neural network is constructed as a global approximation operator. Taking a single-layer feedforward neural network as an example, if the input , output , then the network structure can be expressed as:
[0057]
[0058] in, is the weight, is the bias, is the activation function. Extend the network to L layers, As input, As the final output, an L-layer neural network can be defined as:
[0059]
[0060] in, , is the number of hidden layers, and the weights in all layers and bias The artificial neural network framework is built on TensorFlow2 and Keras, and background noise function tomography is achieved by constructing a deep neural network with a specific number of layers and neurons, or by combining multiple different deep neural networks.
[0061] Next, a loss function with physical constraints will be constructed. Because travel time represented solely by a general artificial neural network cannot guarantee that the output model conforms to the physical laws of seismic wave propagation, it is necessary to add physical constraints to the neural network (i.e., a physical information neural network). This invention uses the eikonal equation as prior information and adds it to the neural network. Using this as a physical constraint, two neural networks, one for travel time and one for velocity, are constructed to output travel time and phase velocity information. This constraint can be expressed as:
[0062]
[0063] in, From the source The seismic waves generated at the receiving point When walking, It is the acceptance point The seismic wave phase velocity at and are the hyperparameters in the travel time neural network and the velocity neural network respectively. Add this physical constraint to the loss function:
[0064]
[0065] in, is the weight coefficient of the physical constraint, is the observed travel time. At the same time, since travel time and velocity cannot be negative, this information is used as a boundary condition for further constraints.
[0066] Next, we will select appropriate hyperparameters, activation functions, and optimizers. We will choose appropriate hyperparameters such as the learning rate, number of iterations, number of layers, and number of neurons per layer based on the amount of data used. We will also need to select appropriate activation functions (including ReLU, sigmoid, and tanh) and optimization algorithms (including adam, SGD, etc.) for inversion. These choices will be based on computational needs, as described in the Keras module.
[0067] S104 , after performing a second preprocessing on the training data, the artificial neural network is trained based on the training data to obtain an initial neural network model, wherein the training data includes the original seismic data and travel time data.
[0068] In the embodiment of the present application, the artificial neural network after the aforementioned steps can be regarded as a constructed physical information neural network. The neural network can be trained by performing the second pre-processed training data to obtain a preliminarily trained initial neural network model.
[0069] In one embodiment, the second preprocessing of the training data includes:
[0070] Convert geographic coordinate system to Cartesian coordinate system based on Mercator projection;
[0071] Standardize and normalize the training data;
[0072] Coordinate system grid division is performed based on the station distance between each station and the data distribution of the input data.
[0073] In the embodiment of the present application, in order to better train the model, in the preprocessing stage, the geographic coordinate system is converted to a Cartesian coordinate system using a Mercator projection, and the training data is standardized and normalized to remove the impact of data units, magnitudes, dimensions, and other factors on the training, thereby improving the convergence speed and accuracy of the model. Finally, the coordinate system grid is divided based on the station distance between each station and the data distribution of the input data to achieve model parameterization.
[0074] S105 , acquiring target seismic data of a target area, processing the target seismic data based on the initial neural network model, and obtaining underground velocity structure information of the target area.
[0075] In the embodiment of the present application, the aforementioned steps are followed to construct and train an initial neural network that meets the requirements. The target seismic data corresponding to the target area for underground structure prediction is input into the initial neural network for processing, and the underground velocity structure information of the target area can be ultimately obtained.
[0076] In one embodiment, the initial neural network model includes a travel time neural network and a speed neural network;
[0077] The step of acquiring target seismic data of a target area and processing the target seismic data based on the initial neural network model to obtain underground velocity structure information of the target area includes:
[0078] Acquire target seismic data of a target area, and process the target seismic data based on the travel time neural network and the velocity neural network to obtain target travel time data and target velocity data;
[0079] The underground velocity structure information of the target area is generated based on the target travel time data and the target velocity data.
[0080] In the embodiments of the present application, considering that the corresponding travel time data and phase velocity data are required to fully generate the underground structure information of the target area, the aforementioned process of adjusting the neural network by adding a loss function containing physical constraints is used to train a travel time neural network and a velocity neural network for travel time data prediction, thereby calculating the target travel time data and target velocity data. Based on the target travel time data and target velocity data, the underground velocity structure information of the target area can be ultimately generated.
[0081] The following will be combined with the Figure 2 , the background noise einofunctional imaging device based on physical information neural network provided by the embodiment of the present application is introduced in detail. Figure 2 The background noise function imaging device based on the physical information neural network shown is used to perform the present application Figure 1 For the convenience of explanation, only the part related to the embodiment of the present application is shown. For the specific technical details not disclosed, please refer to the present application. Figure 1 The embodiment shown.
[0082] See Figure 2 , Figure 2 This is a structural diagram of a background noise einofunctional imaging device based on a physical information neural network provided by an embodiment of the present application. Figure 2 As shown, the device includes:
[0083] An acquisition module 201 is configured to acquire original seismic data and perform a first preprocessing on the original seismic data;
[0084] A cross-correlation module 202 is configured to cross-correlate background noise in the original seismic data to obtain travel time data between at least one set of station pairs;
[0085] A construction module 203 is used to construct an artificial neural network, construct a loss function containing physical constraints based on the eikonal equation, and add the loss function to the artificial neural network;
[0086] A training module 204 is configured to perform a second preprocessing on the training data and then train the artificial neural network based on the training data to obtain an initial neural network model, wherein the training data includes the original seismic data and travel time data;
[0087] The prediction module 205 is used to obtain target seismic data of the target area, process the target seismic data based on the initial neural network model, and obtain underground velocity structure information of the target area.
[0088] In one embodiment, the acquisition module 201 includes:
[0089] The first preprocessing unit is used to sequentially perform format conversion, remove mean, remove instrument response, resample data, and filter the original seismic data.
[0090] In one embodiment, the cross-correlation module 202 includes:
[0091] The cross-correlation unit is used to divide the original seismic data based on frequency bands, and perform cross-correlation on the background noise in the original seismic data of each frequency band to obtain travel time data between at least one set of station pairs.
[0092] In one embodiment, the training module 204 includes:
[0093] A conversion unit for converting a geographic coordinate system into a Cartesian coordinate system based on the Mercator projection;
[0094] A processing unit, used for standardizing and normalizing the training data;
[0095] The division unit is used to divide the coordinate system into grids based on the station distances between the stations and the data distribution of the input data.
[0096] In one embodiment, the prediction module 205 includes:
[0097] an acquisition unit, configured to acquire target seismic data of a target area, and process the target seismic data based on the travel time neural network and the velocity neural network to obtain target travel time data and target velocity data;
[0098] A generating unit is configured to generate underground velocity structure information of the target area based on the target travel time data and the target velocity data.
[0099] Those skilled in the art will clearly understand that the technical solutions of the embodiments of the present application can be implemented with the help of software and / or hardware. "Unit" and "module" in this specification refer to software and / or hardware that can independently perform or cooperate with other components to perform specific functions, where the hardware can be, for example, a field-programmable gate array (FPGA) or an integrated circuit (IC).
[0100] Each processing unit and / or module in the embodiments of the present application may be implemented by an analog circuit that implements the functions described in the embodiments of the present application, or may be implemented by software that executes the functions described in the embodiments of the present application.
[0101] See also Figure 3 , which shows a schematic diagram of the structure of an electronic device involved in an embodiment of the present application, the electronic device can be used to implement Figure 1 The method in the embodiment shown. Figure 3 As shown, the electronic device 300 may include: at least one central processor 301 , at least one network interface 304 , a user interface 303 , a memory 305 , and at least one communication bus 302 .
[0102] The communication bus 302 is used to implement the connection and communication between these components.
[0103] The user interface 303 may include a display screen (Display) and a camera (Camera). Optionally, the user interface 303 may also include a standard wired interface and a wireless interface.
[0104] The network interface 304 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).
[0105] The central processing unit 301 may include one or more processing cores. The central processing unit 301 utilizes various interfaces and circuits to connect various components within the electronic device 300. It executes instructions, programs, code sets, or instruction sets stored in the memory 305 and accesses data stored in the memory 305 to perform various functions and process data for the terminal 300. Optionally, the central processing unit 301 may be implemented in hardware using at least one of a digital signal processing (DSP), a field-programmable gate array (FPGA), and a programmable logic array (PLA). The central processing unit 301 may integrate one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. The CPU primarily processes the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing content displayed on the display; and the modem handles wireless communications. It is understood that the modem may also be implemented as a separate chip, rather than integrated into the central processing unit 301.
[0106] Among them, the memory 305 may include a random access memory (RAM) or a read-only memory (Read-Only Memory). Optionally, the memory 305 includes a non-transitory computer-readable storage medium. The memory 305 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 305 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 305 may also be optionally at least one storage device located away from the aforementioned central processor 301. As Figure 3 As shown, the memory 305 as a computer storage medium may include an operating system, a network communication module, a user interface module, and program instructions.
[0107] exist Figure 3In the electronic device 300 shown, the user interface 303 is mainly used to provide an input interface for the user and obtain user input data; and the central processing unit 301 can be used to call the background noise function imaging application based on physical information neural network stored in the memory 305 and specifically perform the following operations:
[0108] Acquiring original seismic data and performing a first preprocessing on the original seismic data;
[0109] Cross-correlating background noise in the raw seismic data to obtain travel time data between at least one set of station pairs;
[0110] Constructing an artificial neural network, constructing a loss function including physical constraints based on the eikonal equation, and adding the loss function to the artificial neural network;
[0111] After performing a second preprocessing on the training data, the artificial neural network is trained based on the training data to obtain an initial neural network model, wherein the training data includes the original seismic data and travel time data;
[0112] Target seismic data of a target area is acquired, and the target seismic data is processed based on the initial neural network model to obtain underground velocity structure information of the target area.
[0113] The present application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above method. The computer-readable storage medium may include, but is not limited to, any type of disk, including a floppy disk, an optical disk, a DVD, a CD-ROM, a microdrive, a magneto-optical disk, a ROM, a RAM, an EPROM, an EEPROM, a DRAM, a VRAM, a flash memory device, a magnetic card or an optical card, a nanosystem (including a molecular memory IC), or any type of medium or device suitable for storing instructions and / or data.
[0114] It should be noted that for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by this application.
[0115] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.
[0116] In the several embodiments provided in this application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interface, and the indirect coupling or communication connection of the device or unit can be electrical or other forms.
[0117] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.
[0118] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.
[0119] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a memory and includes a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned memory includes: U disk, read-only memory (ROM), random access memory (RAM), mobile hard disk, magnetic disk, or optical disk, etc., various media that can store program code.
[0120] Those skilled in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be completed by instructing related hardware through a program. The program may be stored in a computer-readable memory, which may include a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.
[0121] The above is only an exemplary embodiment of the present disclosure and cannot be used to limit the scope of the present disclosure. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the specification and practicing the disclosure herein, those skilled in the art will easily think of the implementation scheme of the present disclosure. This application is intended to cover any variation, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the art that are not recorded in the present disclosure. The description and examples are to be regarded as exemplary only, and the scope and spirit of the present disclosure are defined by the claims.
Claims
1. A background noise eicofunctional imaging method based on physical information neural network, characterized in that: The method comprises: Acquiring original seismic data and performing a first preprocessing on the original seismic data; Cross-correlating background noise in the raw seismic data to obtain travel time data between at least one set of station pairs; Constructing an artificial neural network, constructing a loss function including physical constraints based on the eikonal equation, and adding the loss function to the artificial neural network; After performing a second preprocessing on the training data, the artificial neural network is trained based on the training data to obtain an initial neural network model, wherein the training data includes the original seismic data and travel time data, and the initial neural network model includes a travel time neural network and a velocity neural network; Acquiring target seismic data of a target area, processing the target seismic data based on the initial neural network model, and obtaining underground velocity structure information of the target area; Among them, physical constraints The calculation formula is: in, From the source The seismic waves generated at the receiving point When walking, It is the receiving point The seismic wave phase velocity at and are the hyperparameters in the travel time neural network and the velocity neural network respectively; Loss Function The calculation formula is: in, is the weight coefficient of the physical constraint, It is the observation time. is the number of earthquake focal points, is the number of receiving points, is the jth earthquake source The seismic wave generated at the i-th receiving point The original travel time, is the jth earthquake source under physical constraints The seismic wave generated at the i-th receiving point When walking, is the i-th receiving point under physical constraints The seismic wave phase velocity at are the original hyperparameters of the travel time neural network.
2. The method according to claim 1, characterized in that The first preprocessing of the original seismic data includes: The raw seismic data are sequentially subjected to format conversion, mean removal, instrument response removal, data resampling, and filtering.
3. The method according to claim 1, characterized in that Cross-correlating the background noise in the original seismic data to obtain travel time data between at least one set of station pairs includes: The original seismic data is divided based on frequency bands, and background noise in the original seismic data of each frequency band is cross-correlated to obtain travel time data between at least one set of station pairs.
4. The method according to claim 1, wherein The second preprocessing of the training data includes: Convert geographic coordinate system to Cartesian coordinate system based on Mercator projection; Standardize and normalize the training data; The coordinate system grid is divided based on the station distance between stations and the data distribution of the input data.
5. The method according to claim 1, characterized in that The step of acquiring target seismic data of a target area and processing the target seismic data based on the initial neural network model to obtain underground velocity structure information of the target area includes: Acquire target seismic data of a target area, and process the target seismic data based on the travel time neural network and the velocity neural network to obtain target travel time data and target velocity data; The underground velocity structure information of the target area is generated based on the target travel time data and the target velocity data.
6. A background noise eicofunctional imaging device based on physical information neural network, characterized in that: The device comprises: an acquisition module, configured to acquire original seismic data and perform a first preprocessing on the original seismic data; A cross-correlation module, configured to cross-correlate background noise in the original seismic data to obtain travel time data between at least one set of station pairs; A construction module, configured to construct an artificial neural network, construct a loss function including physical constraints based on the eikonal equation, and add the loss function to the artificial neural network; A training module is configured to perform a second preprocessing on the training data and then train the artificial neural network based on the training data to obtain an initial neural network model, wherein the training data includes the original seismic data and travel time data, and the initial neural network model includes a travel time neural network and a velocity neural network; A prediction module, configured to obtain target seismic data of a target area, process the target seismic data based on the initial neural network model, and obtain underground velocity structure information of the target area; Among them, physical constraints The calculation formula is: in, From the source The seismic waves generated at the receiving point When walking, It is the receiving point The seismic wave phase velocity at and are the hyperparameters in the travel time neural network and the velocity neural network respectively; Loss Function The calculation formula is: in, is the weight coefficient of the physical constraint, It is the observation time. is the number of earthquake focal points, is the number of receiving points, is the jth earthquake source The seismic wave generated at the i-th receiving point The original travel time, is the jth earthquake source under physical constraints The seismic wave generated at the i-th receiving point When walking, is the i-th receiving point under physical constraints The seismic wave phase velocity at are the original hyperparameters of the travel time neural network.
7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.
8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.