Method, system, device and storage medium for constructing a multiple wave sample library

Abstract

The present invention discloses a method, system, device and storage medium for constructing a multiple wave sample library. The method includes obtaining marine formation data of a target sea area, and determining a marine layered geological model including a seawater layer geological model according to the marine formation data; obtaining forward simulation data; the forward simulation data is obtained according to a preset forward simulation observation system; based on preset boundary conditions, the forward simulation data is respectively input into the marine layered geological model and the seawater layer geological model for calculation to obtain first seismic record data and second seismic record data; calculating based on the first seismic record data and the second seismic record data to determine a multiple wave sample data set; performing sample augmentation and normalization processing on the multiple wave sample data set to determine the multiple wave sample library of the target sea area. The embodiments of the present invention can improve the overall accuracy of the constructed free surface multiple wave sample library and can be widely applied to the field of seismic exploration technology.

Description

Technical Field

[0001] The present invention relates to the technical field of seismic exploration, and particularly relates to a method, a system, a device and a storage medium for constructing a multiple wave sample library. Background Art

[0002] The free surface multiple wave in marine seismic exploration refers to the wave formed by the seismic wave reflecting or refracting at the interface between seawater and the seabed and then propagating back and forth multiple times in the ocean. In marine seismic exploration, constructing a free surface multiple wave sample library in marine seismic exploration means collecting and sorting out multiple wave data under different sea areas and different geological conditions, and establishing a database containing information such as the propagation characteristics of multiple waves, reflection and refraction patterns, etc. These sample data usually include the propagation characteristics of free surface multiple waves, velocity models, and the mutual relationships of different waveforms. Constructing such a sample library is of great significance: First, it can provide standardized reference data for marine seismic exploration, helping seismologists accurately distinguish multiple waves from primary waves and reducing the interference of multiple waves on the imaging of underground structures; Second, by comparing multiple wave samples in different scenarios, exploration personnel can optimize the solution process of the wave equation and improve the processing accuracy and efficiency of marine exploration data.

[0003] In the prior art, a certain amount of multiple wave sample libraries can be generated by seismic forward modeling methods, but there are still certain differences between the data obtained by forward modeling and the actual data, and the data processing of some multiple wave samples is not comprehensive, and direct waves and other interferences are still retained, resulting in low overall accuracy of the constructed free surface multiple wave sample library. Summary of the Invention

[0004] In view of this, to solve one of the above problems, an object of an embodiment of the present invention is to provide a method, a system, a device and a storage medium for constructing a multiple wave sample library, which can improve the overall accuracy of the constructed free surface multiple wave sample library.

[0005] On the one hand, an embodiment of the present invention provides a method for constructing a multiple wave sample library, including the following steps:

[0006] Obtain the marine formation data of the target sea area, and determine the marine layered geological model according to the marine formation data; the marine layered geological model includes the geological model of the seawater layer;

[0007] Obtain forward modeling data; the forward modeling data is obtained according to a preset forward modeling observation system;

[0008] Based on preset boundary conditions, input the forward modeling data into the marine layered geological model for calculation to obtain first seismic record data; based on preset boundary conditions, input the forward modeling data into the geological model of the seawater layer for calculation to obtain second seismic record data;

[0009] Based on the first seismic record data and the second seismic record data, perform calculations to determine a multiple wave sample data set;

[0010] Based on the multiple wave sample data set, perform sample augmentation and normalization processing to determine a multiple wave sample library for the target sea area.

[0011] Specifically, the marine formation data includes seawater layer data and seabed formation data. Determining a marine layered geological model based on the marine formation data includes:

[0012] Based on the seawater layer data, determine seawater layer model construction parameters; the seawater layer model construction parameters include seawater layer depth, number of seawater layers, and seawater layer width;

[0013] Based on the seabed formation data, determine seabed formation construction parameters; the seabed formation model construction parameters include terrain parameters, geological parameters, and physical parameters;

[0014] Based on the seawater layer construction parameters, construct a seawater layer model, based on the seabed formation model construction parameters, construct a seabed formation model, and based on the seawater layer model and the seabed formation model, determine the marine layered geological model.

[0015] Specifically, the forward simulation observation system includes a seismic source set and geophones; the forward simulation observation system is constructed in the following manner:

[0016] Obtain shot gather setting parameters; the shot gather setting parameters include the number of shots, shot spacing, and trace spacing;

[0017] Construct a seismic source set according to the shot gather setting parameters; the seismic source set is used to simulate the occurrence of an earthquake;

[0018] Based on the shot gather design parameters, determine geophone setting parameters; the geophone setting parameters include the arrangement pattern;

[0019] Construct a geophone set according to the geophone setting parameters; the geophones are used to collect forward simulation data;

[0020] Based on the seismic source set and the geophone set, construct the forward simulation observation system.

[0021] Specifically, the preset boundary conditions include free boundary conditions and absorbing boundary conditions; based on the preset boundary conditions, input the forward simulation data into the marine layered geological model for calculation to obtain the first seismic record data, including:

[0022] Perform a first setting on the marine layered geological model based on free boundary conditions, input the forward simulation data into the marine layered geological model after the first setting for calculation, and obtain the first seismic record data under free boundary conditions;

[0023] Perform a second setting on the marine layer geological model based on absorbing boundary conditions, input the forward simulation data into the marine layered geological model after the second setting for calculation, and obtain the first seismic record data under absorbing boundary conditions;

[0024] Obtain the first seismic record data based on the first seismic record data under free boundary conditions and the first seismic record data under absorbing boundary conditions.

[0025] Specifically, the second seismic record data includes the second seismic record data under free boundary conditions and the second seismic record data under absorbing boundary conditions; performing calculations based on the first seismic record data and the second seismic record data to obtain a multiple wave sample data set, including:

[0026] Performing calculations based on the first seismic record data and the second seismic record data under free boundary conditions to obtain multiple wave sample data containing free surface multiples;

[0027] Performing calculations based on the first seismic record data and the second seismic record data under absorbing boundary conditions to obtain multiple wave sample data with free surface multiples removed;

[0028] Obtain a multiple wave sample data set based on the multiple wave sample data containing free surface multiples and the multiple wave sample data with free surface multiples removed.

[0029] Specifically, performing sample augmentation and normalization processing based on the multiple wave sample data set to determine the multiple wave sample library of the target sea area, including:

[0030] Performing sample augmentation on the multiple wave sample data containing free surface multiples, and obtaining input data according to the results of sample augmentation;

[0031] Performing normalization processing on the input data and the multiple wave sample data with free surface multiples removed respectively, and constructing the multiple wave sample database of the target sea area according to the results of normalization processing.

[0032] Specifically, the performing sample augmentation on the multiple wave sample data containing free surface multiples includes:

[0033] Performing flipping on the multiple wave sample data containing free surface multiples; the flipping includes left - right flipping or front - back flipping;

[0034] And / or, adding noise to the multiple wave sample data containing free surface multiples based on a preset range and a preset signal-to-noise ratio;

[0035] And / or, adding gather missing to the multiple wave sample data containing free surface multiples based on a preset ratio; the gather missing includes regular missing and random missing.

[0036] On the other hand, an embodiment of the present invention further provides a construction system for a multiple wave sample library, including:

[0037] A first module, configured to obtain marine formation data of a target sea area, and determine a marine layered geological model according to the marine formation data; the marine layered geological model includes a seawater layer geological model;

[0038] A second module, configured to obtain forward modeling data; the forward modeling data is obtained according to a preset forward modeling observation system;

[0039] A third module, configured to input the forward modeling data into the marine layered geological model for calculation based on a preset boundary condition to obtain first seismic record data; input the forward modeling data into the seawater layer geological model for calculation based on a preset boundary condition to obtain second seismic record data;

[0040] A fourth module, configured to calculate based on the first seismic record data and the second seismic record data to determine a multiple wave sample data set;

[0041] A fifth module, configured to perform sample augmentation and normalization processing based on the multiple wave sample data set to determine the multiple wave sample library of the target sea area.

[0042] On the other hand, an embodiment of the present invention further provides a construction device for a multiple wave sample library, including:

[0043] At least one processor;

[0044] At least one memory, configured to store at least one program;

[0045] When the at least one program is executed by the at least one processor, the at least one processor implements the method as described above.

[0046] On the other hand, an embodiment of the present invention further provides a computer-readable storage medium, in which a processor-executable program is stored, and the processor-executable program is used to execute the method as described above when executed by a processor.

[0047] Implementing the embodiments of the present invention includes the following beneficial effects: In this embodiment, by obtaining the marine formation data of the target water area, a marine layered geological model including a seawater layer model is established. Under preset conditions, the forward simulation data is respectively input into the seawater layer model and the overall marine layered geological model to calculate the corresponding seismic record data. Based on the obtained seismic record data, calculations, sample augmentation, and normalization processing are performed to obtain a multiple wave sample database. On the one hand, the multiple wave samples obtained by mutual calculation of the seismic record data of the seawater layer and the overall marine situation can effectively eliminate the influence of direct waves and seawater layer interference, improving the accuracy of the multiple wave sample data. On the other hand, through sample augmentation of the sample data, the sample data can be made more in line with the actual situation, reducing the difference between the sample data and the actual data. And normalizing the sample data can unify the form of the sample data, making the obtained sample data more integral. From these two aspects, the overall accuracy of the constructed multiple wave sample library can be effectively improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 FIG. is a schematic flow chart of the steps of a method for constructing a multiple wave sample library provided by an embodiment of the present invention;

[0049] Figure 2 FIG. is a structural block diagram of a marine layered geological model provided by this embodiment;

[0050] Figure 3 FIG. is a structural block diagram of a seawater layer model provided by this embodiment;

[0051] Figure 4 FIG. is a structural block diagram of a forward simulation observation system provided by an embodiment of the present invention;

[0052] Figure 5 FIG. is a schematic flow chart of the process for obtaining multiple wave sample data provided by an embodiment of the present invention;

[0053] Figure 6 FIG. is a schematic flow chart of a sample augmentation method provided by an embodiment of the present invention;

[0054] Figure 7 FIG. is a schematic flow chart of another sample augmentation method provided by an embodiment of the present invention;

[0055] Figure 8 FIG. is a schematic flow chart of a method for processing missing rules of sample data provided by an embodiment of the present invention;

[0056] Figure 9 FIG. is a schematic flow chart of a method for randomly processing missing sample data provided by an embodiment of the present invention;

[0057] Figure 10It is a schematic flow chart of a method for normalizing sample data provided by an embodiment of the present invention;

[0058] Figure 11 It is a schematic step flow chart of another method for constructing a multiple wave sample library provided by an embodiment of the present invention;

[0059] Figure 12 It is an example diagram of several marine geological and geophysical models provided by an embodiment of the present invention;

[0060] Figure 13 It is another example diagram of several marine geological and geophysical models provided by an embodiment of the present invention;

[0061] Figure 14 It is an example diagram of several constructed marine seismic multiple wave sample libraries provided by an embodiment of the present invention;

[0062] Figure 15 It is a structural block diagram of a system for constructing a multiple wave sample library provided by an embodiment of the present invention;

[0063] Figure 16 It is a structural block diagram of a device for constructing a multiple wave sample library provided by an embodiment of the present invention. Specific embodiments

[0064] The following further elaborates on the present invention in detail in conjunction with the accompanying drawings and specific embodiments. For the step numbers in the following embodiments, they are only set for the convenience of elaboration and explanation, and no limitation is imposed on the order between steps. The execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

[0065] Explanations for several terms involved in this application are as follows:

[0066] Tesseral: A numerical modeling software widely used in seismic exploration, mainly used for simulating seismic wave propagation under complex geological structures. It uses the finite difference method (FDM) to solve the wave equation and can simulate phenomena such as seismic wave reflection, refraction, and transmission. Tesseral has high flexibility and can handle various complex subsurface media, including non-uniform and anisotropic situations. In addition, it can also perform processing and analysis of multiple waves, providing important support for oil and gas exploration, mineral resource exploration, and environmental seismic monitoring.

[0067] Finite difference method (FDM): A numerical analysis method widely used for solving partial differential equations, especially suitable for simulating physical phenomena such as wave propagation and heat conduction. In FDM, continuous time and space are discretized into finite grid points, and derivatives are approximated by difference formulas, thereby transforming the differential equation into an algebraic equation for solution.

[0068] Seismic forward modeling: A technique that uses numerical methods to simulate the propagation process of seismic waves in subsurface media. By inputting the physical parameters of the subsurface structure (such as velocity, density, etc.) and wave source information, forward modeling can predict the propagation modes of seismic waves in different media, including phenomena such as reflection, refraction, and transmission. This process helps to study subsurface structures, identify the relationship between seismic waves and subsurface structures, and provide theoretical support for seismic exploration.

[0069] Boundary conditions: Refer to the constraint conditions defined on the boundary of the problem domain in mathematical and physical problems. They are used to limit the solution of a physical problem within a specific region and play a decisive role in solving the problem. When solving partial differential equations, boundary conditions help to determine the behavior of the system. For example, in heat conduction, fluid dynamics, and seismic wave propagation, boundary conditions can describe specific values or variation laws of temperature, velocity, or pressure.

[0070] Free boundary conditions: In problems such as seismic wave propagation and fluid mechanics, free boundary conditions mean that there are no fixed values or specific derivative constraints on the boundary, and they are usually set to the natural state of the external environment. For example, in seismic forward modeling, free boundary conditions may mean that seismic waves are not forced to reflect or absorb at the outer boundary of the subsurface medium but propagate naturally. It is often used to simulate the boundaries of open systems and can reflect the natural boundary characteristics in actual physical phenomena.

[0071] Absorbing boundary conditions: In the numerical simulation of seismic wave, acoustic wave, or light wave propagation, absorbing boundary conditions can effectively reduce the influence of boundary effects, making the wave disappear in an "absorbing" way at the boundary, thus simulating the wave propagation behavior in open space. Common absorbing boundary conditions include the perfectly matched layer (PML); this method can reduce the error caused by boundary reflection, improve the authenticity and stability of the numerical solution, and is widely used in seismic forward modeling, acoustic simulation, and electromagnetic wave propagation and other fields.

[0072] Perfectly Matched Layer (PML): A technique of absorbing boundary conditions used in the numerical simulation of wave propagation, aiming to effectively reduce boundary reflection and simulate the wave propagation behavior in open space. PML sets a layer of virtual absorbing medium at the boundary of the computational domain, making the wave almost completely disappear when it touches the boundary, avoiding the unnatural reflection caused by traditional boundary conditions. It is widely used in the numerical simulation of problems such as seismic waves, acoustic waves, and electromagnetic waves, and can significantly improve the calculation accuracy and stability. The design of PML absorbs wave energy without introducing too much computational complexity, becoming an important tool in fields such as seismic forward modeling, acoustic simulation, and electromagnetic wave propagation direct wave.

[0073] Seawater Layer Reflection: It refers to the reflection phenomenon of sound waves or other types of waves at the ocean surface or the interface between different underwater media. Due to the density and sound speed differences between seawater and air, submarine rock formations, or underwater objects, part of the wave will be reflected when the wave encounters an interface during propagation. Especially in the propagation of sound waves, seawater surface reflection will have an important impact on underwater acoustic detection (such as underwater acoustic positioning and sonar systems).

[0074] Acoustic Impedance: It refers to the impedance degree of a medium to sound waves when sound waves propagate in the medium. It is the product of the density and sound speed of the medium, usually represented by "Z", and the unit is Pascal-second per meter (Pa·s / m). The magnitude of acoustic impedance determines the propagation characteristics of sound waves in the medium. When sound waves propagate from one medium to another, if the acoustic impedance differences between the two media are large, reflection and refraction phenomena of sound waves will occur.

[0075] Data Augmentation: A commonly used data preprocessing technique in machine learning and deep learning. By performing various transformation operations on the original data, new samples are generated to expand the training dataset. Common augmentation methods include rotation, translation, cropping, flipping, color change, etc. of images, or synonym replacement, sentence rearrangement, etc. of texts. Data augmentation can effectively alleviate the shortage of training data, avoid overfitting, and improve the generalization ability and robustness of the model.

[0076] Sample Rotation: A data augmentation technique widely used in the fields of machine learning and computer vision, especially in image processing and deep learning. By rotating the original samples, samples at different angles can be generated, thus increasing the diversity of training data and improving the generalization ability of the model. Sample rotation not only helps the model to handle rotational invariance, but also prevents overfitting and enhances the adaptability of the model in different scenarios.

[0077] Noise Injection: A technique commonly used in machine learning and deep learning. By adding noise to the training data, the robustness and generalization ability of the model are enhanced. The noise is usually randomly generated and can be added in different forms, such as Gaussian noise, salt-and-pepper noise, etc. Noise injection helps to prevent the model from overfitting to the training data because it forces the model to learn more general features when processing inputs with uncertainties. Through appropriate noise interference, the model can better adapt to unknown inputs and improve its performance and stability in practical applications.

[0078] Feature Missing: Refers to the situation of introducing partial feature missing in the dataset to simulate the scenario of incomplete data or partial information loss in reality. This method is usually used to test and evaluate the performance of machine learning models when facing incomplete data. By artificially adding feature missing, researchers can observe how the model processes these missing values and adopt appropriate strategies (such as imputation, deletion, or inference of missing features) to improve the robustness of the model.

[0079] Rule Missing: Refers to removing or ignoring some key rules or information in specific samples of the dataset to simulate the possible situation of missing data in the real world. This method is usually used to train machine learning models to improve their adaptability to data missing or incomplete situations. By artificially adding rule missing, the model can learn how to cope with and process missing information, thus enhancing its robustness in practical applications and avoiding performance degradation when facing incomplete data.

[0080] Random Missingness: Refers to randomly selecting and removing some values in certain sample features of the dataset to simulate the situation of data loss. This method is usually used to test the performance of machine learning models in a data missing environment. The introduction of random missingness can help evaluate the robustness of the model when dealing with incomplete data. In this way, the model can learn how to cope with missing information and thus improve its adaptability in practical applications.

[0081] Convolutional Neural Network (CNN): A deep learning model widely used in fields such as image recognition, video analysis, and natural language processing. Its core idea is to extract local features in the input data by simulating the human visual system using convolutional operations. A CNN usually consists of multiple convolutional layers, pooling layers, and fully connected layers. The convolutional layer performs local perception on the data through convolutional kernels, the pooling layer is used to reduce the data dimension and prevent overfitting, and the fully connected layer is responsible for the final classification or regression task. Through the multi-layer structure, the CNN can automatically learn and extract complex features in the data, significantly improving the effects of tasks such as image processing and speech recognition.

[0082] Normalization: A data preprocessing method that aims to convert feature data of different scales and units into the same range. Common normalization methods include minimum-maximum normalization and Z-score standardization. Minimum-maximum normalization scales the data to a specified interval (such as 0 to 1) through linear transformation, while Z-score standardization makes the data present a standard normal distribution by subtracting the mean and dividing by the standard deviation. Normalization helps to speed up the model training process, improve the convergence speed of the model, and avoid certain features from having too much impact on the model due to scale differences. The effect is particularly significant when using optimization algorithms such as gradient descent.

[0083] like Figure 1 As shown, an embodiment of the present invention provides a method for constructing a multiple wave sample library, which includes steps S100 to S500 as shown below.

[0084] S100: Acquire marine stratum data of the target sea area, and determine a marine layered geological model according to the marine stratum data; the marine layered geological model includes a seawater layer geological model.

[0085] The marine stratigraphic data of the target sea area is obtained, including the seawater layer depth, stratigraphic conditions, scale, etc., and then a marine layered geological model is constructed based on the marine stratigraphic data, wherein the seawater layer geological model can be separated from the marine layered geological model.

[0086] Optionally, the modeling process can be modeled through the Tesserial full-wavefield forward simulation observation system, mainly referring to the existing marine stratum model, by performing a series of overlapping polygons in the Tesserial full-wavefield forward simulation observation system (setting the width and depth of each layer), and in the process of model creation, using reflection layers, velocity, density, anisotropy and other parameters from the acoustic logging curve to establish different marine geological models (mainly using the acoustic wave velocity parameters, in this embodiment, the acoustic wave propagation velocity of the seawater layer is set to 1500m / s).

[0087] Furthermore, through modeling using the Tesserial full-wavefield forward simulation observation system, multiple categories of marine geological models (different seawater depths, different formation conditions, and different scales) can be constructed to simulate sample data under different seabed formation conditions.

[0088] Specifically, the marine stratum data includes seawater layer data and seabed stratum data, and in step S100, determining the marine layered geological model according to the marine stratum data includes:

[0089] S110: Determine seawater layer model construction parameters based on the seawater layer data; the seawater layer model construction parameters include seawater layer depth, seawater layer number and seawater layer width.

[0090] Determine the construction parameters for constructing the seawater layer model based on the seawater layer data of the target sea area; among them, the construction parameters include the depth, quantity, width, etc. of the seawater layer.

[0091] S120: Determine the seabed formation construction parameters based on the seabed formation data; the seabed formation model construction parameters include topographic parameters, geological parameters, and physical parameters.

[0092] Determine the construction parameters for constructing the seabed formation model based on the seabed formation data of the target water area; among them, the construction parameters include topographic parameters, geological parameters, and physical parameters, etc. of the seabed formation.

[0093] S130: Construct the seawater layer model based on the seawater layer construction parameters, construct the seabed formation model based on the seabed formation model construction parameters, and determine the marine layered geological model based on the seawater layer model and the seabed formation model.

[0094] Construct the seawater layer model and the seabed formation model respectively through the corresponding construction parameters, and determine the overall marine layered geological model based on the constructed seawater layer model and seabed formation model. At the same time, the seawater layer model can be separated from the constructed overall marine layered geological model; as Figure 2 and Figure 3 shown, Figure 2 is a structural block diagram of a marine layered geological model provided in this embodiment (the horizontal axis Z represents the width, the vertical axis X represents the depth; the side schematic diagram is for sonic reference, and Acousticvelocity represents the sonic velocity), Figure 3 is a structural block diagram of a seawater layer model provided in this embodiment, and this seawater layer model is separated from the marine layered geological model (the horizontal axis Z represents the width, the vertical axis X represents the depth; the side schematic diagram is for sonic reference, and Acoustic velocity represents the sonic velocity).

[0095] S200: Obtain the forward simulation data; the forward simulation data is obtained according to the preset forward simulation observation system.

[0096] By constructing the forward simulation observation system, obtain the forward simulation data for generating the sample data.

[0097] Specifically, in seismic forward modeling, by setting different boundary conditions, single-shot seismic records that simulate with / without free-surface multiples can be obtained. However, in marine seismic exploration, seismic data is often collected using a streamer. At the same time, the single-shot seismic data in direct forward modeling often has direct waves, which will cause certain interference to deep learning training if not removed. Although it can be removed by seismic data processing and excision means, it is rather cumbersome. In the embodiments of the present invention, we design the forward modeling acquisition system with reference to the actual marine seismic exploration process, adopt the one-sided shot method, with geophones for reception, and at the same time collect data by moving the geophones along with the shot source to restore the actual marine seismic exploration process.

[0098] Specifically, the forward modeling acquisition system includes a source set and geophones. The forward modeling acquisition system in step S200 is constructed in the following manner:

[0099] Obtain the shot gather setting parameters; the shot gather setting parameters include the number of shots, shot interval, and trace interval;

[0100] Construct a source set according to the shot gather setting parameters; the source set is used to simulate the occurrence of an earthquake;

[0101] Determine the geophone setting parameters based on the shot gather design parameters; the geophone setting parameters include the array pattern;

[0102] Construct a geophone set according to the geophone setting parameters; the geophones are used to collect forward modeling data;

[0103] Construct a forward modeling acquisition system based on the source set and the geophone set.

[0104] Specifically, by setting the relevant parameters of the shot gather, determine the relevant parameters of the source set and the geophone set and construct them respectively, and construct a forward modeling acquisition system through the constructed source set and geophone set. Further, the forward modeling acquisition system design adopts a multiple-fold acquisition system. At the same time, the source position is at the end of the geophone array. By setting a certain number of shots, shot interval, and trace interval, and moving the geophone array along with the source, seismic forward modeling is carried out to obtain synthetic seismic shot gather data, as Figure 4 shown, which is a structural block diagram of a forward modeling acquisition system provided by an embodiment of the present invention.

[0105] S300: Based on preset boundary conditions, input the forward modeling data into the marine layered geological model for calculation to obtain the first seismic record data; based on preset boundary conditions, input the forward modeling data into the seawater layer geological model for calculation to obtain the second seismic record data.

[0106] Using preset boundary conditions, the marine layered geological model and the seawater layer geological model are respectively set. After the setting is completed, the forward simulation data are respectively input into the corresponding models for calculation to obtain the corresponding seismic record data, which serves as the data basis for generating the multiple wave samples.

[0107] Specifically, the preset boundary conditions include free boundary conditions and absorbing boundary conditions; in step S300, based on the preset boundary conditions, the forward simulation data are input into the marine layered geological model for calculation to obtain the first seismic record data, including:

[0108] Perform the first setting on the marine layered geological model based on the free boundary conditions, input the forward simulation data into the marine layered geological model after the first setting for calculation, and obtain the first seismic record data under the free boundary conditions;

[0109] Perform the second setting on the marine layered geological model based on the absorbing boundary conditions, input the forward simulation data into the marine layered geological model after the second setting for calculation, and obtain the first seismic record data under the absorbing boundary conditions;

[0110] Obtain the first seismic record data based on the first seismic record data under the free boundary conditions and the first seismic record data under the absorbing boundary conditions.

[0111] Specifically, in seismic forward simulation, the present invention uses a mature finite-difference acoustic wave forward simulation method to generate synthetic data, and respectively sets and calculates the marine layered geological model based on the free boundary conditions and the absorbing boundary conditions to obtain the seismic record data (the first seismic record data) of the marine layered geological model under the free boundary conditions and the absorbing boundary conditions; among them, the seismic record data under the free boundary conditions has free surface multiples, while the seismic record data under the absorbing boundary conditions has no free surface multiples.

[0112] Similarly, through the above calculation process, the seismic record data (the second seismic record data) of the seawater layer model under the free boundary conditions and the absorbing boundary conditions can be obtained.

[0113] S400: Calculate based on the first seismic record data and the second seismic record data to determine the multiple wave sample data set.

[0114] The first seismic record data corresponds to the seismic record data of the overall marine layered geological model, and the second seismic record data corresponds to the seismic record data of the seawater layer model. There are the same direct waves and seawater layer reflections and other interference effects in the seismic record data corresponding to the two models. By performing mutual calculations on the two seismic record models, the interference effects such as direct waves and seawater layer reflections can be eliminated, and multiple wave seismic data without direct waves and seawater layer reflections can be obtained under the preset boundary conditions.

[0115] Specifically, the second seismic recording data includes the second seismic recording data under free boundary conditions and the second seismic recording data under absorbing boundary conditions; in step S400, calculations are performed based on the first seismic recording data and the second seismic recording data to obtain a multiple wave sample data set, including:

[0116] Calculations are performed based on the first seismic recording data and the second seismic recording data under free boundary conditions to obtain multiple wave sample data including free surface multiples;

[0117] Calculations are performed based on the first seismic recording data and the second seismic recording data under absorbing boundary conditions to obtain multiple wave sample data with free surface multiples removed;

[0118] Based on the multiple wave sample data including free surface multiples and the multiple wave sample data with free surface multiples removed, a multiple wave sample data set is obtained.

[0119] Under different boundary conditions, the seismic recording data of the overall ocean layered geological model and the seismic recording data of the seawater layer model are mutually calculated (subtraction calculation) to eliminate the interference effects of direct waves and seawater layer reflections, etc., and multiple wave sample data under different conditions are obtained. Among them, the multiple wave sample data including free surface multiples is obtained by mutual calculation under free boundary conditions; the multiple wave sample data with free surface multiples removed is obtained by mutual calculation under absorbing boundary conditions. Summarize them, use the sample data including free surface multiples as input data, and the sample data with free surface multiples removed as output data, and summarize to obtain a multiple wave sample data set.

[0120] As Figure 5 shown, it is a schematic flowchart of a process for obtaining multiple wave sample data provided by an embodiment of the present invention. Specifically, Figure 5 in the figures (a1) to (a3) show the process of obtaining multiple wave sample data including free surface multiples, where Figure 5 in the figures (a1) to (a3) are respectively the seismic recording data of the ocean layered geological model under free boundary conditions, the seismic recording data of the seawater layer model under free boundary conditions, and the multiple wave sample data including free surface multiples; Figure 5 in the figures (b1) to (b3) show the process of obtaining multiple wave sample data with free surface multiples removed, where Figure 5 in the figures (b1) to (b3) are respectively the seismic recording data of the ocean layered geological model under absorbing boundary conditions, the seismic recording data of the seawater layer model under absorbing boundary conditions, and the multiple wave sample data with free surface multiples removed.

[0121] S500: Augment and normalize the multi-path sample data set to determine the multi-path sample library of the target sea area.

[0122] By augmenting the obtained multi-path sample data set, the reliability of the data is improved; secondly, the data after sample augmentation is normalized to achieve unified size, serving as the data basis for the composition of the sample library.

[0123] Specifically, augmenting and normalizing the multi-path sample data set to determine the multi-path sample library of the target sea area includes:

[0124] S510: Augment the multi-path sample data containing free-surface multiples, and obtain input data according to the result of sample augmentation.

[0125] By augmenting the obtained multi-path sample data set, the reliability of the data is improved to obtain input data. Among them, after performing the above operations on the established marine geological model, multi-path sample data containing / removing free-surface multiples after removing direct waves and seawater layer reflections can be obtained. The seismic data obtained through forward modeling often has good quality, which is somewhat different from the data collected during actual marine seismic exploration. At the same time, there are also certain limitations in the quantity of data generated by forward modeling. In order to further expand the quantity of samples and obtain seismic records similar to actual marine seismic exploration, the embodiments of the present invention adopt methods of sample rotation, noise addition, and trace gather deletion in sequence for sample augmentation.

[0126] Specifically, augmenting the multi-path sample data containing free-surface multiples includes:

[0127] S511: Flip the multi-path sample data containing free-surface multiples; the flipping includes left-right flipping or front-back flipping.

[0128] In the sample augmentation of deep learning image recognition, rotating image data at any angle is a commonly used sample augmentation method. Combining with the single-shot data in actual marine seismic exploration, in the present invention, left-right / up-down flipping of the sample data is adopted to increase the quantity and variety of training data. As Figure 6 shown, Figure 6 is a schematic flowchart of a sample augmentation method provided by the embodiments of the present invention. Among them, Figure 6 Figure (6a) in it is the original sample data, and respectively refer to left-right flipping and up-down flipping of the original sample data, Figure 6 Figures (6b) and (6c) in it are the sample data obtained by left-right flipping and the sample data obtained by flipping respectively.

[0129] S512: And / or, adding noise to the multiple wave sample data containing free surface multiples based on a preset range and a preset signal-to-noise ratio.

[0130] In the actual marine seismic exploration, the data acquisition process will be interfered by various types of noise. In the present invention, by randomly adding noise with a certain signal-to-noise ratio within a certain range to simulate the seismic exploration data in the real marine environment, the noise addition process in the forward modeling data can be expressed as:

[0131]

[0132] Wherein, is the marine seismic data with noise, is the synthetic marine seismic data obtained by forward modeling, is the added random noise.

[0133] Specifically, as shown in Figure 7 shown, Figure 7 is a schematic flowchart of another sample augmentation method provided by an embodiment of the present invention; Figure 7 Figures (7a)-(7c) in

[0134] S513: And / or, adding gather missing to the multiple wave sample data containing free surface multiples; the gather missing includes regular missing and random missing.

[0135] Receiving gather random / regular missing, when the number of lateral marine seismic streamers is small, it will cause insufficient spatial sampling rate of the collected data, resulting in a phenomenon similar to regularized data missing. In addition, when a geophone is damaged or the signal is interfered, there will be randomly missing traces in the collected seismic data. In view of this situation, the present invention performs a certain proportion of regular missing and random missing on the multiple wave sample data respectively to cover as much data as possible in the actual marine seismic exploration.

[0136] Specifically, as shown in Figure 8 and Figure 9 shown, Figure 8 is a schematic flowchart of a method for processing regular missing of sample data provided by an embodiment of the present invention, wherein Figure 8 Figures (d1)-(d3), Figures (e1)-(e3) and Figures (f1)-(f3) in and column correspond to three instances of the process of performing regular missing on the original multiple wave sample data, and column and degree Results of rule missing processing; Figure 9 It is a schematic flowchart of a method for performing random missing processing on sample data provided by an embodiment of the present invention. Figure 9 In FIGS. (g1)-(g3), FIGS. (h1)-(h3), and FIGS. (i1)-(i3), they are respectively three instances of the process of performing random missing on sample data, listed and column corresponds to the result of performing random missing processing on the original multiple wave sample data to a certain degree and degree of the result.

[0137] It should be noted that the above sample augmentation process changes the rotation direction, noise level, and the number of missing traces of the sample data containing free surface multiple waves, while the corresponding removal of free surface multiple wave sample data does not change.

[0138] S520: Normalize the input data and the multiple wave sample data after removing free surface multiple waves respectively, and construct a multiple wave sample database of the target sea area according to the results of the normalization process.

[0139] Normalize the data after sample augmentation to achieve unified size, serving as the data basis for the composition of the sample library.

[0140] Specifically, the purpose of obtaining the multiple wave sample database is to use it as the training data basis for a seismic survey model based on a neural network. Among them, the multiple wave sample data containing free surface multiple waves is used as the input data sample, and the multiple wave sample data after removing free surface multiple waves is used as the output data sample (output label). Therefore, for the convenience of deep learning, when inputting the data into a seismic survey model based on a neural network, the embodiment of the present invention normalizes the multiple wave sample data and adjusts the sample size as needed, and finally forms a standardized multiple wave sample database. As Figure 10 shown, Figure 10 It is a schematic flowchart of a method for performing normalization processing on sample data provided by an embodiment of the present invention. Among them, Figure 10 FIG. (j1) in it is the input sample, Figure 10 FIGS. (j2) and (j3) in it are sample data normalized to 1024×1024 and 512×512 respectively; Figure 10 FIG. (k1) in it is the input sample, Figure 10 FIGS. (k2) and (k3) in it are sample data normalized to 1024×1024 and 512×512 respectively.

[0141] In summary, the embodiments of the present method at least include the following beneficial effects:

[0142] (1) The method of obtaining synthetic seismic data by forward modeling. Compared with actual seismic data, synthetic seismic data can fit any theoretically existing marine geological model data, and the types and quantities of data can be controlled through model design and observation systems, and the corresponding seismic data labels for removing marine seismic multiples are more accurate.

[0143] (2) In the present invention, by subtracting seismic data sets from the marine layered geological model and its corresponding seawater layer model, the obtained marine seismic records not only have no interference from direct waves, but also remove the seawater layer reflection information that is generally not required in marine seismic exploration, making the generated sample set more focused on suppressing free-surface multiples in marine seismic exploration.

[0144] (3) In sample augmentation, sample augmentation close to marine seismic exploration is adopted instead of the conventional deep learning sample augmentation method, and the data after sample augmentation is closer to the data collected in the actual marine seismic exploration process.

[0145] Such as Figure 11 shown, Figure 11 is a schematic flow chart of the steps of another method for constructing a multiple wave sample library provided by an embodiment of the present invention. The embodiment of the present invention also provides another method for constructing a multiple wave sample library, including the following steps (1) - (8):

[0146] (1) Establishment of various types of marine geological and geophysical sound velocity models containing seawater layers and separation of seawater layer sound velocities;

[0147] (2) Design of a forward modeling observation system that conforms to marine seismic exploration;

[0148] (3) Use the finite difference method to set the free surface and absorption surface for the two models respectively (set the seawater surface as a perfectly matched layer absorption boundary condition), and perform four forward modeling simulations;

[0149] (4) Remove seawater layer reflections and direct waves by subtracting the corresponding seawater layer forward modeled seismic reflection data from the forward modeled seismic reflection data of the marine geological and geophysical model;

[0150] (5) The seismic data after removing seawater layer reflections and direct waves are successively subjected to sample flipping → adding noise with different signal-to-noise ratios → randomly / regularly missing traces for sample augmentation;

[0151] (6) Normalize and adjust the size of the seismic data after sample augmentation;

[0152] (7) Correlate the seismic data containing free surface multiples with the seismic data without free surface multiples;

[0153] (8) Repeat steps (1) → (8) until sufficient seismic data containing free-surface multiples and their corresponding seismic data without free-surface multiples are generated. Combine the seismic data generated by all models to form an ocean seismic free-surface multiple sample library.

[0154] Specifically, in the implementation process of this method embodiment, a total of 12 different marine geological and geophysical models were established (as shown in Figure 12 and Figure 12 and Figure 13 are example diagrams of several marine geological and geophysical models provided by the embodiments of the present invention. The horizontal axis s represents the width, and the vertical axis V represents the depth). The size and seawater layer thickness of each marine geological model are different. At the same time, the observation system design is based on an ocean seismic streamer. The trace interval, shot interval, and number of shots are all different. Different main frequencies were used for acquisition during the forward modeling process. After forward modeling and simulation of 12 different marine geological and geophysical models and their corresponding seawater layers, the seismic records with direct waves and seawater layer reflections removed were obtained by subtracting the forward modeling results of the seawater layer. Then, sample augmentation was performed through sample flipping → adding noise with different signal-to-noise ratios → randomly / regularly missing traces in the gather. Finally, normalization was carried out. The normalized size of the seismic records was set to 256×256. Finally, 8500 groups of sufficient seismic data containing free-surface multiples and their corresponding single-shot seismic records without free-surface multiples were generated to form an ocean seismic free-surface multiple sample library, as shown in Figure 14 and Figure 14 is an example diagram of several constructed ocean seismic multiple sample libraries provided by the embodiments of the present invention. Among them, the example diagram in row is the input sample, and the example diagram in row is the output label.

[0155] In summary, the method embodiment of the present invention has at least the following beneficial effects:

[0156] (1) Single-shot data for marine seismic exploration without direct wave and seawater layer reflection interference. Currently, in some methods for generating single-shot seismic data of ocean seismic free-surface multiples through forward modeling simulation, the seismic data obtained from forward modeling simulation are often directly input into the seismic exploration model for training and learning, without removing the direct waves and seawater layer reflections in the seismic data. These two will affect the training of the deep learning model in the form of noise interference. In the present invention, forward modeling and simulation are respectively performed on the seawater layer models corresponding to the constructed marine geological models, and the interference of direct waves is removed by subtracting the seismic data sets to obtain single-shot data for marine seismic exploration without direct wave and seawater layer reflection interference.

[0157] (2) The forward simulation observation system design conforms to marine seismic exploration. In the current forward simulation process of some marine seismic exploration, the observation system design uses an observation system where the source is in the middle and the detectors are on both sides for reception. However, in actual marine seismic exploration, a streamer design is used for actual seismic data acquisition, resulting in differences between the forward simulation and the actual collected data. Therefore, in the observation system design of the marine seismic forward simulation process of the present invention, the source is on the side of the detector, and the detector moves with the source, so as to maximize the restoration of the data acquisition method during the marine seismic exploration process and reduce the difference between the simulated value and the actual value.

[0158] (3) Sample augmentation close to actual marine seismic data. In the actual acquisition of marine seismic exploration data, various unexpected situations will occur, and at the same time, it is faced with various noise interferences. The noise interference will cause a low signal-to-noise ratio in the actual marine seismic data, and the damage of the detector will also cause the missing of traces in the seismic record. However, the data of the forward simulation is often of good quality, and there is a large difference from the actually collected data. In order to establish marine seismic data close to the actual situation, the present invention selects a certain sample augmentation method to increase the training set data while being close to the actual marine seismic single-shot record. The sample augmentation includes a realistic 90° rotation, left and right flipping to increase the quantity and variety, and also includes the addition of noise with a certain signal-to-noise ratio, random trace missing and regular trace missing to restore the real marine seismic single-shot acquisition data.

[0159] (4) The process of corresponding training samples with low labeling cost and accuracy to actual labels. During the deep learning training process, it is necessary to label the input seismic single-shot record containing the marine seismic free-surface multiple waves to the corresponding seismic single-shot record without the marine seismic free-surface multiple waves. Marking by the traditional method through manual processing is time-consuming, laborious and costly, and cannot completely suppress the multiple waves. The present invention uses the method of marine forward simulation. Theoretically, through different boundary condition settings and other same forward simulation settings, accurate and corresponding seismic single-shot records containing / without the marine seismic free-surface multiple waves are fitted, and thus different types of seismic single-shot data samples containing the free-surface multiple waves as the input and their corresponding label data without the free-surface multiple waves as the output can be obtained.

[0160] As Figure 15 shown, the embodiment of the present invention also provides a system for constructing a multiple-wave sample library, including:

[0161] The first module is used to obtain the marine formation data of the target sea area and determine the marine layered geological model according to the marine formation data; the marine layered geological model includes the seawater layer geological model;

[0162] The second module is used to obtain the forward simulation data; the forward simulation data is obtained according to the preset forward simulation observation system;

[0163] A third module, configured to input forward modeling data into an ocean layered geological model for calculation based on preset boundary conditions to obtain first seismic recording data; and input forward modeling data into a seawater layer geological model for calculation based on preset boundary conditions to obtain second seismic recording data;

[0164] A fourth module, configured to calculate based on the first seismic recording data and the second seismic recording data to determine a multiple wave sample data set;

[0165] A fifth module, configured to perform data preprocessing based on the multiple wave sample data set and construct a multiple wave sample library for the target sea area according to the result of the data preprocessing.

[0166] It can be seen that the content in the above method embodiments is applicable to the system embodiments herein. The functions specifically implemented by the system embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those of the above method embodiments.

[0167] As Figure 16 shown, an embodiment of the present invention further provides a device for constructing a multiple wave sample library, including:

[0168] At least one processor;

[0169] At least one memory, configured to store at least one program;

[0170] When the at least one program is executed by the at least one processor, the at least one processor implements the method for constructing a multiple wave sample library as described in the above method embodiments.

[0171] Among them, the memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. The memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state storage devices. In some embodiments, the memory may optionally include a remote memory remotely disposed relative to the processor, and these remote memories may be connected to the processor through a network. Examples of the above network include but are not limited to the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0172] It can be seen that the content in the above method embodiments is applicable to the device embodiments herein. The functions specifically implemented by the device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those of the above method embodiments.

[0173] In addition, an embodiment of the present application also discloses a computer program product or a computer program, which is stored in a computer-readable storage medium. The processor of the computer device can read the computer program from the computer-readable storage medium, and the processor executes the computer program, so that the computer device executes the method for constructing a multiple wave sample library described in the above method embodiment.

[0174] An embodiment of the present invention also provides a computer-readable storage medium, which stores a program executable by a processor. The program executable by the processor is used to implement the above method when executed by the processor. Similarly, the content in the above method embodiment is applicable to this storage medium embodiment. The functions specifically implemented by this storage medium embodiment are the same as those in the above method embodiment, and the beneficial effects achieved are also the same as those in the above method embodiment.

[0175] It can be understood that all or some of the steps and systems in the methods disclosed above can be implemented as software, firmware, hardware, and their appropriate combinations. Some physical components or all physical components can be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or can be implemented as hardware, or can be implemented as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, which can include a computer storage medium (or a non-transitory medium) and a communication medium (or a transitory medium). As is well known to those of ordinary skill in the art, computer storage media include volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information, such as computer-readable instructions, data structures, program modules, or other data. Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory, or other memory technologies, CD-ROM, digital versatile disk (DVD), or other optical disk storage, magnetic cassette, tape, magnetic disk storage, or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer. In addition, as is well known to those of ordinary skill in the art, communication media typically contain computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transmission mechanism, and can include any information delivery medium.

[0176] The above is a specific description of the preferred embodiments of the present invention, but the present invention is not limited to the described embodiments. Those skilled in the art can make various equivalent deformations or substitutions without departing from the spirit of the present invention, and these equivalent deformations or substitutions are all included in the scope defined by the claims of the present application.

Claims

1. A method for constructing a multiple wave sample library, characterized in that, Including: Obtaining marine formation data of a target sea area, and determining a marine layered geological model according to the marine formation data; the marine layered geological model includes a seawater layer geological model; Obtaining forward simulation data; the forward simulation data is obtained according to a preset forward simulation observation system; Based on preset boundary conditions, inputting the forward simulation data into the marine layered geological model for calculation to obtain first seismic record data; Based on preset boundary conditions, inputting the forward simulation data into the seawater layer geological model for calculation to obtain second seismic record data; the preset boundary conditions include free boundary conditions and absorbing boundary conditions; the first seismic record data includes first seismic record data under free boundary conditions and first seismic record data under absorbing boundary conditions; the second seismic record data includes second seismic record data under free boundary conditions and second seismic record data under absorbing boundary conditions; Calculating based on the first seismic record data under the free boundary conditions and the second seismic record data under the free boundary conditions to obtain multi-wave sample data including free surface multiples; calculating based on the first seismic record data under the absorbing boundary conditions and the second seismic record data under the absorbing boundary conditions to obtain multi-wave sample data with free surface multiples removed; based on the multi-wave sample data including free surface multiples and the multi-wave sample data with free surface multiples removed, obtaining a multi-wave sample data set; Performing sample augmentation and normalization processing based on the multi-wave sample data set to determine a multi-wave sample library of the target sea area.

2. The method according to claim 1, characterized in that The marine formation data includes seawater layer data and seabed formation data, and the determining of the marine layered geological model according to the marine formation data includes: Determining seawater layer model construction parameters based on the seawater layer data; the seawater layer model construction parameters include seawater layer depth, number of seawater layers, and seawater layer width; Determining seabed formation model construction parameters based on the seabed formation data; the seabed formation model construction parameters include topographic parameters, geological parameters, and physical parameters; Constructing a seawater layer model based on the seawater layer model construction parameters, constructing a seabed formation model based on the seabed formation model construction parameters, and determining the marine layered geological model based on the seawater layer model and the seabed formation model.

3. The method according to claim 1, characterized in that The forward simulation observation system includes a source set and geophones; the forward simulation observation system is constructed in the following manner: Obtaining shot gather setting parameters; the shot gather setting parameters include the number of shots, shot spacing, and trace spacing; Constructing a source set according to the shot gather setting parameters; the source set is used to simulate the occurrence of an earthquake; Determining geophone setting parameters based on the shot gather design parameters; the geophone setting parameters include the arrangement pattern; Constructing a geophone set according to the geophone setting parameters; the geophones are used to collect forward simulation data; Constructing the forward simulation observation system based on the source set and the geophone set.

4. The method according to claim 1, wherein The inputting of the forward simulation data into the marine layered geological model for calculation based on preset boundary conditions to obtain first seismic record data includes: Perform a first setting on the marine layered geological model based on free boundary conditions, input the forward simulation data into the marine layered geological model after the first setting for calculation, and obtain the first seismic record data under free boundary conditions; Perform a second setting on the marine layer geological model based on absorbing boundary conditions, input the forward simulation data into the marine layered geological model after the second setting for calculation, and obtain the first seismic record data under absorbing boundary conditions; Obtain the first seismic record data based on the first seismic record data under free boundary conditions and the first seismic record data under absorbing boundary conditions.

5. The method according to claim 1, wherein The sample augmentation and normalization processing based on the multiple wave sample data set to determine the multiple wave sample library of the target sea area includes: Perform sample augmentation on the multiple wave sample data including free surface multiples, and obtain input data according to the result of sample augmentation; Perform normalization processing on the input data and the multiple wave sample data after removing free surface multiples respectively, and construct the multiple wave sample database of the target sea area according to the result of normalization processing.

6. The method according to claim 5, characterized in that, The performing sample augmentation on the multiple wave sample data including free surface multiples includes: Perform flipping on the multiple wave sample data including free surface multiples; the flipping includes left - right flipping or front - back flipping; And / or, add noise to the multiple wave sample data including free surface multiples based on a preset range and preset signal - to - noise ratio; And / or, add trace - gather missing data to the multiple wave sample data including free surface multiples based on a preset ratio; the trace - gather missing includes regular missing and random missing.

7. A system for constructing a multiple wave sample library, characterized in that Includes: A first module for obtaining marine stratum data of the target sea area and determining a marine layered geological model according to the marine stratum data; the marine layered geological model includes a seawater layer geological model; A second module for obtaining forward simulation data; the forward simulation data is obtained according to a preset forward simulation observation system; A third module for inputting the forward simulation data into the marine layered geological model for calculation based on preset boundary conditions to obtain the first seismic record data; Input the forward simulation data into the seawater layer geological model for calculation based on preset boundary conditions to obtain the second seismic record data; the preset boundary conditions include free boundary conditions and absorbing boundary conditions; the first seismic record data includes the first seismic record data under free boundary conditions and the first seismic record data under absorbing boundary conditions; the second seismic record data includes the second seismic record data under free boundary conditions and the second seismic record data under absorbing boundary conditions; A fourth module, configured to perform calculations based on the first seismic recording data under the free boundary condition and the second seismic recording data under the free boundary condition to obtain multiple wave sample data including free surface multiples; perform calculations based on the first seismic recording data under the absorption boundary condition and the second seismic recording data under the absorption boundary condition to obtain multiple wave sample data with free surface multiples removed; and obtain a multiple wave sample data set based on the multiple wave sample data including free surface multiples and the multiple wave sample data with free surface multiples removed. A fifth module, configured to perform sample augmentation and normalization processing based on the multiple wave sample data set to determine a multiple wave sample library for the target sea area.

8. An apparatus for constructing a multiple wave sample library, characterized in that Comprising: At least one processor; At least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements the method according to any one of claims 1 to 6.

9. A computer-readable storage medium storing a program executable by a processor, characterized in that, The program executable by the processor, when executed by the processor, is used to execute the method according to any one of claims 1 to 6.

Images