Elastic wave full-waveform inversion method, system and device and medium

Abstract

The invention relates to the technical field of elastic waves, and discloses an elastic wave full-waveform inversion method, which comprises the steps of inputting observed seismic data into a preset neural network encoder, and extracting to obtain high-dimensional feature representation; inputting the high-dimensional feature representation into a preset neural network decoder to obtain a gradient update value; obtaining an elastic parameter model based on the gradient update value and the initial elastic parameter model; obtaining Lame parameters based on the longitudinal wave velocity, the transverse wave velocity and the medium density; based on a Lame parameter and an elastic wave equation, obtaining synthetic seismic data; obtaining an objective function based on the observed seismic data and the synthesized seismic data; using the target function to iteratively update the preset neural network weight until the loss function reaches a first preset threshold value or the number of iterations reaches a second preset threshold value, and obtaining a target gradient update value; and based on the target gradient update value and the target elastic parameter model, obtaining a final target elastic parameter model so as to relieve discomfort caused by simultaneous inversion of multiple elastic parameters.

Description

Technical Field

[0001] The present application relates to the field of elastic wave technology, and in particular to an elastic wave full waveform inversion method, system, device and medium. Background Art

[0002] With the continuous advancement of oil and gas exploration and development technologies, building high-precision subsurface parameter models can provide reliable elastic parameters for subsequent oil and gas interpretation, which is of great significance to seismic imaging, geological interpretation, and oil and gas exploration and development. Full waveform inversion (FWI) has become a hot research topic due to its ability to fully utilize the kinematic and dynamic information of seismic waves. Deep learning methods are gradually being introduced into the elastic wave FWI process and have become a common technical approach.

[0003] However, traditional acoustic full-waveform inversion (FWI) is often ill-posed and prone to cycle skipping. Furthermore, the joint inversion of multiple elastic parameters (P-wave velocity, S-wave velocity, and medium density) and multi-component data processing in elastic FWI further exacerbate this ill-posedness, making it difficult to obtain a satisfactory inversion parameter model. In particular, when the initial elastic parameter model is biased, the inversion process is prone to cycle skipping, making high-precision modeling difficult. The inversion process often suffers from strong ill-posedness and severe parameter gradient coupling. Summary of the Invention

[0004] The purpose of the embodiments of the present application is to provide a full-waveform inversion method, system, device and medium for elastic wave, so as to solve the problems in the prior art where the inversion process is very prone to frequency hopping, is difficult to achieve high-precision modeling, and often faces problems such as strong instability and severe parameter gradient coupling.

[0005] To achieve the above objectives, the present application provides, in a first aspect, a method for elastic wave full waveform inversion, the method comprising: Acquiring initial synthetic seismic data, observed seismic data, and a first initial elastic parameter model, wherein the observed seismic data includes a horizontal component of an initial particle velocity vector and a vertical component of an initial particle velocity vector; Inputting the observed seismic data into an encoder module of a preset neural network to extract high-dimensional feature representations of the observed seismic data, wherein the preset neural network is trained using a Unet model; The high-dimensional feature representation is input into the decoder module of the preset neural network to obtain the initial gradient update value of the elasticity parameter; Based on the initial gradient update value and the first initial elastic parameter model, a first target elastic parameter model related to the longitudinal wave velocity, the shear wave velocity and the medium density is obtained; Based on the longitudinal wave velocity, shear wave velocity and medium density, the target medium Lamé parameters are obtained; Based on the target medium Lame parameter and the elastic wave equation, synthetic multi-component seismic data is obtained, wherein the synthetic multi-component seismic data includes a horizontal component of the target particle velocity vector and a vertical component of the target particle velocity vector; Based on the observed seismic data and the synthetic multi-component seismic data, an objective function is obtained; Iteratively updating the weights of a preset neural network using an objective function until an error obtained based on the observed seismic data, the synthesized multi-component seismic data, and the loss function reaches a first preset threshold or the number of iterative updates reaches a second preset threshold, thereby obtaining a target gradient update value of the elastic parameter; Based on the target gradient update value and the second target elastic parameter model, a final target elastic parameter model related to the target longitudinal wave velocity, target shear wave velocity and target medium density is obtained, wherein the second target elastic parameter model is determined based on the gradient update value and the elastic parameter model obtained by iteratively updating the weights of the previous preset neural network.

[0006] In the embodiment of the present application, the first target elastic parameter model includes:

[0007] Where, represents the first objective elastic parameter model, represents the decoder module, Indicates the encoder module, represents the network weight, represents the observed earthquake data, Indicates custom weights, represents the update step size, Represents the first initial elastic parameter model.

[0008] In the embodiment of the present application, the target medium Lamé parameters are obtained based on the longitudinal wave velocity, the shear wave velocity and the medium density, including: Based on the longitudinal wave velocity, shear wave velocity, medium density and the Lamé parameter calculation formula, the target medium Lamé parameter is obtained. The Lamé parameter calculation formula includes:

[0009]

[0010] Where, and Both represent the target medium Lamé parameters, represents the density of the medium, represents the shear wave velocity, represents the longitudinal wave velocity.

[0011] In the embodiment of the present application, obtaining initial synthetic seismic data includes: Obtaining initial medium density, initial longitudinal wave velocity and initial shear wave velocity; Based on the initial medium density, initial longitudinal wave velocity and initial shear wave velocity, the initial medium Lamé parameters are obtained; Based on the initial medium Lame parameters and elastic wave equation, the initial synthetic seismic data are obtained.

[0012] In the embodiment of the present application, the elastic wave equation includes:

[0013]

[0014]

[0015]

[0016]

[0017] Where, represents the density of the medium, Indicates the horizontal plane, represents the vertical surface, represents the horizontal component of the particle velocity vector, represents the seismic wave propagation time, express Stress response in the direction, express Stress response in the direction, represents the vertical component of the particle velocity vector, express Stress response in the direction, represents the medium Lamé parameter, Indicates the medium Lamé parameter.

[0018] In an embodiment of the present application, the encoder module includes an encoder, which is composed of multiple 3*3 convolutional layers and 2*2 maximum pooling layers alternating.

[0019] In an embodiment of the present application, the decoder module includes a first decoder, a second decoder and a third decoder. The first decoder is used to generate longitudinal wave velocity, the second decoder is used to generate shear wave velocity, and the third decoder is used to generate medium density. The first decoder, the second decoder and the third decoder are all composed of multiple 2*2 deconvolution layers.

[0020] A second aspect of the present application provides an elastic wave full waveform inversion system, the system comprising: an acquisition module, configured to acquire initial synthetic seismic data, observed seismic data, and a first initial elastic parameter model, wherein the observed seismic data includes a horizontal component of an initial particle velocity vector and a vertical component of an initial particle velocity vector; An extraction module is used to input the observed seismic data into an encoder module of a preset neural network to extract a high-dimensional feature representation of the observed seismic data, wherein the preset neural network is trained by a Unet model; The first obtaining module is used to input the high-dimensional feature representation into the decoder module of the preset neural network to obtain the initial gradient update value of the elastic parameter; A second obtaining module is configured to obtain a first target elastic parameter model related to the longitudinal wave velocity, the shear wave velocity, and the medium density based on the initial gradient update value and the first initial elastic parameter model; A third obtaining module is used to obtain the target medium Lame parameters based on the longitudinal wave velocity, the shear wave velocity and the medium density; a fourth obtaining module, configured to obtain synthetic multi-component seismic data based on the target medium Lame parameter and the elastic wave equation, wherein the synthetic multi-component seismic data includes a horizontal component of the target particle velocity vector and a vertical component of the target particle velocity vector; a fifth obtaining module, for obtaining a target function based on the observed seismic data and the synthesized multi-component seismic data; an iterative update module for iteratively updating the weights of a preset neural network using an objective function until an error obtained based on the observed seismic data, the synthesized multi-component seismic data, and the loss function reaches a first preset threshold or the number of iterative updates reaches a second preset threshold, thereby obtaining a target gradient update value of the elastic parameter; The sixth obtaining module is used to obtain a final target elastic parameter model related to the target longitudinal wave velocity, target shear wave velocity and target medium density based on the target gradient update value and the second target elastic parameter model, wherein the second target elastic parameter model is determined based on the gradient update value and the elastic parameter model obtained by iteratively updating the weights of the previous preset neural network.

[0021] A third aspect of the present application provides an elastic wave full waveform inversion device, comprising: a memory configured to store instructions; The processor is configured to call instructions from the memory and implement the elastic wave full waveform inversion method according to the first aspect when executing the instructions.

[0022] A fourth aspect of the present application provides a machine-readable storage medium having stored thereon instructions for causing a machine to execute the elastic wave full waveform inversion method according to the first aspect described above.

[0023] The above technical solution effectively alleviates the problem of local artifacts in the results caused by the limited expression ability of the neural network and the problem of instability caused by gradient crosstalk between different elastic parameters in the joint inversion of multiple elastic parameters.

[0024] Other features and advantages of the embodiments of the present application will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the following detailed description, they are used to explain the embodiments of the present application but do not constitute a limitation on the embodiments of the present application. In the accompanying drawings: Figure 1 A schematic diagram of a flow chart of an elastic wave full waveform inversion method according to an embodiment of the present application is shown; Figure 2 Schematically shows a schematic diagram of a real elastic parameter model based on a two-dimensional elastic Marmousi model simulation according to an embodiment of the present application; Figure 3 A schematic diagram of an initial elastic parameter model based on the present application according to an embodiment of the present application is schematically shown; Figure 4 A schematic diagram of an elastic parameter model obtained based on the elastic wave full waveform inversion method of the present application according to an embodiment of the present application is schematically shown; Figure 5 A schematic diagram of an elastic parameter model obtained by inversion using a traditional physics-driven elastic wave full waveform inversion method according to an embodiment of the present application is shown; Figure 6 Schematically shows a comparison diagram of a single-channel curve of elastic parameters according to an embodiment of the present application; Figure 7 A schematic diagram showing a comparison of the change trend of a normalized loss function according to an embodiment of the present application is schematically shown; Figure 8 Schematically illustrates a schematic diagram of a real elastic parameter model simulated based on a schematic diagram of a two-dimensional elastic BP model according to an embodiment of the present application; Figure 9 A schematic diagram of another initial elastic parameter model according to an embodiment of the present application is schematically shown; Figure 10 A schematic diagram of another elastic parameter model obtained based on the elastic wave full waveform inversion method of the present application according to an embodiment of the present application is schematically shown; Figure 11 A schematic diagram of an elastic parameter model obtained by inverting another elastic wave full waveform inversion method based on traditional physical drive according to an embodiment of the present application is shown; Figure 12 Fig. 6 schematically shows a comparison diagram of another normalized loss function trend according to an embodiment of the present application. DETAILED DESCRIPTION

[0026] To make the objectives, technical solutions, and superiorities of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. It should be understood that the specific embodiments described herein are merely used to explain and illustrate the embodiments of the present application, and should not be used to limit the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort fall within the scope of protection of the present application.

[0027] It should be noted that the acquisition, transmission, storage, use, processing, and the like of data in the technical solutions of the present application comply with relevant provisions of laws and regulations. In the embodiments of the present application, some industry existing solutions, components, models, and the like may be mentioned, which should be considered as exemplary, and the purpose is merely to illustrate the feasibility in the implementation of the technical solutions of the present application, but does not mean that the applicant has or will necessarily use the solutions.

[0028] It should be noted that if the embodiments of the present application involve directional indications (such as up, down, left, right, front, back, and the like), the directional indications are merely used to explain the relative positional relationship, motion condition, and the like between components in a certain specific posture (as shown in the drawings), and if the specific posture changes, the directional indications also change accordingly.

[0029] In addition, if the embodiments of the present application involve descriptions of “first”, “second”, and the like, the descriptions of “first”, “second”, and the like are merely for description purposes, and should not be understood as indicating or implying the relative importance of the indicated technical features or implicitly indicating the number of the indicated technical features. Therefore, the features limited by “first”, “second” can explicitly or implicitly include at least one of the features. In addition, the technical solutions of the various embodiments can be combined with each other, but must be based on the fact that a person of ordinary skill in the art can implement the combination, and when the combination of the technical solutions contradicts each other or cannot be implemented, it should be considered that the combination of the technical solutions does not exist, and is not within the scope of protection claimed by the present application.

[0030] Figure 1 Fig. 1 schematically shows a flowchart of an elastic wave full waveform inversion method according to an embodiment of the present application. As shown in Fig. 1, the elastic wave full waveform inversion method can include the following steps: Figure 1 The elastic wave full waveform inversion method provided by the embodiments of the present application can include the following steps: Step S110: acquiring initial synthetic seismic data, observed seismic data, and a first initial elastic parameter model, wherein the observed seismic data includes a horizontal component of an initial particle velocity vector and a vertical component of an initial particle velocity vector; Step S120: inputting the observed seismic data into an encoder module of a preset neural network to extract high-dimensional feature representations of the observed seismic data, wherein the preset neural network is trained by a Unet model; Step S130: inputting the high-dimensional feature representation into a decoder module of a preset neural network to obtain an initial gradient update value of the elasticity parameter; Step S140: obtaining a first target elastic parameter model related to the longitudinal wave velocity, the shear wave velocity, and the medium density based on the initial gradient update value and the first initial elastic parameter model; Step S150: obtaining the target medium Lamé parameters based on the longitudinal wave velocity, the shear wave velocity and the medium density; Step S160: obtaining synthetic multi-component seismic data based on the target medium Lame parameter and the elastic wave equation, wherein the synthetic multi-component seismic data includes a horizontal component of the target particle velocity vector and a vertical component of the target particle velocity vector; Step S170: obtaining an objective function based on the observed seismic data and the synthesized multi-component seismic data; Step S180: Iteratively updating the weights of the preset neural network using the objective function until the error obtained based on the observed seismic data, the synthesized multi-component seismic data, and the loss function reaches a first preset threshold or the number of iterative updates reaches a second preset threshold, thereby obtaining a target gradient update value of the elastic parameter; Step S190: Based on the target gradient update value and the second target elastic parameter model, a final target elastic parameter model related to the target longitudinal wave velocity, target shear wave velocity and target medium density is obtained, wherein the second target elastic parameter model is determined based on the gradient update value and the elastic parameter model obtained by iteratively updating the weights of the previous preset neural network.

[0031] In step S110, the initial synthetic seismic data is seismic wave data artificially generated by a theoretical model or numerical simulation method, and the observed seismic data is seismic wave data acquired by an actual seismograph. The first initial elastic parameter model is a mathematical framework used in seismology to describe the elastic properties of a medium. The core of the model is to quantify the response of the medium to elastic waves through a set of physical parameters.

[0032] In step S120, the observed seismic data is input into an encoder module of a preset neural network trained by a Unet model to extract a high-dimensional feature representation of the observed seismic data.

[0033] In step S130, the high-dimensional feature representation of the observed seismic data is input into a decoder module of a preset neural network obtained by training the Unet model to obtain an initial gradient update value of the elastic parameter.

[0034] In step S140, based on the initial gradient update value of the elastic parameter obtained in step S130 and the first initial elastic parameter model obtained in step S110, a first target elastic parameter model related to the longitudinal wave velocity, the shear wave velocity, and the medium density is obtained. The longitudinal wave velocity, the shear wave velocity, and the medium density are all elastic parameters.

[0035] In step S150, the target medium Lamé parameters are obtained by converting the longitudinal wave velocity, the shear wave velocity and the medium density.

[0036] In step S160, the target medium Lamé parameters are substituted into the elastic wave equation for calculation to obtain synthetic multi-component seismic data.

[0037] In step S170, in conventional elastic wave full waveform inversion, the objective function typically uses the L2 norm to measure the residual between the observed seismic data and the forward modeled data. However, the present embodiment, taking into account that the elastic wave field contains multiple synthetic multi-component seismic data, calculates the objective function based on the observed seismic data, the synthetic multi-component seismic data, and the conventional L2 norm. The objective function calculation formula includes:

[0038] represents the objective function, represents the vertical component of the particle velocity vector of the synthetic multi-component seismic data, represents the vertical component of the particle velocity vector of the synthetic multi-component seismic data, represents the vertical component of the initial particle velocity vector of the observed seismic data, represents the horizontal component of the initial particle velocity vector of the observed seismic data, represents the spatial coordinates of the detection point, Indicates the maximum sampling time.

[0039] In step S180, the weights of the preset neural network are iteratively updated using the objective function, and the error is calculated using the selected loss function. The formula for calculating the error using the loss function includes:

[0040] In the above formula, represents the number of earthquake sources, represents the observed earthquake data, Represents synthetic multi-component seismic data.

[0041] The weights of the preset neural network are iteratively updated until the error in the loss function calculation reaches a first preset threshold or the number of iterative updates reaches a second preset threshold. At this point, updating the weights of the preset neural network is suspended to obtain a target gradient update value for the elasticity parameter. Illustratively, the first preset threshold and the second preset threshold can be set as needed and are not specifically limited in this embodiment.

[0042] In step S190, based on the target gradient update value and the second target elastic parameter model, a final target elastic parameter model related to the target longitudinal wave velocity, target shear wave velocity and target medium density is obtained, wherein the second target elastic parameter model is determined based on the gradient update value and the elastic parameter model obtained by the iterative update of the weights of the previous preset neural network.

[0043] The embodiment of the present application avoids directly predicting the elastic parameter model as in the prior art by learning the nonlinear mapping relationship between synthetic multi-component seismic data and the elastic parameter update gradient, thereby effectively alleviating the problem of local artifacts in the results caused by the limited expression ability of the neural network and the instability problem caused by gradient crosstalk between different elastic parameters during the joint inversion of multiple elastic parameters (compressional wave velocity, shear wave velocity and medium density).

[0044] In an optional implementation, the first target elastic parameter model includes:

[0045] Where, represents the first objective elastic parameter model, represents the decoder module, Indicates the encoder module, represents the network weight, represents the observed earthquake data, Indicates custom weights, represents the update step size, Represents the first initial elastic parameter model, where the custom weight is defined based on an empirical formula.

[0046] In an optional embodiment, the encoder module includes an encoder, which is composed of multiple 3*3 convolutional layers and 2*2 maximum pooling layers alternating.

[0047] In an embodiment of the present application, in the preset neural network obtained by training the Unet model, the encoder adopts a structure of alternating 3*3 convolutional layers and 2*2 maximum pooling layers to more efficiently capture the high-dimensional feature representation of the observed seismic data.

[0048] In an optional embodiment, the decoder module includes a first decoder, a second decoder and a third decoder, the first decoder is used to generate longitudinal wave velocity, the second decoder is used to generate shear wave velocity, and the third decoder is used to generate medium density, and the first decoder, the second decoder and the third decoder are all composed of multiple 2*2 deconvolution layers.

[0049] In the embodiment of the present application, in the preset neural network obtained by training the Unet model, a multi-decoder structure is used to process the inversion of different elastic parameters separately, which can maintain the consistency of high-dimensional feature representation extraction and realize independent decoding of different elastic parameter gradients, effectively reducing the gradient crosstalk effect between multiple elastic parameters, thereby significantly improving the resolution and robustness of the inversion process. Among them, the specific form of the first target elastic parameter model includes:

[0050]

[0051]

[0052]

[0053] In the above formula, represents the first decoder, represents the second decoder, represents the third decoder, represents the initial longitudinal wave velocity, represents the initial shear wave velocity, represents the initial medium density, represents the target longitudinal wave velocity, represents the target shear wave velocity, Indicates the target medium density.

[0054] In an optional implementation, step S110 may include the following steps: Step S111: obtaining initial medium density, initial longitudinal wave velocity, and initial shear wave velocity; Step S112: obtaining initial medium Lamé parameters based on the initial medium density, initial longitudinal wave velocity, and initial shear wave velocity; Step S113: Obtain initial synthetic seismic data based on the initial medium Lamé parameters and the elastic wave equation.

[0055] In step S111 to step S112, the initial medium Lamé parameters are obtained based on the obtained initial medium density, initial longitudinal wave velocity, initial shear wave velocity and the Lamé parameter calculation formula. The Lamé parameter calculation formula includes:

[0056]

[0057] Where, and Both represent the initial medium Lamé parameters, represents the initial medium density, represents the initial shear wave velocity, represents the initial longitudinal wave velocity.

[0058] In step S113, the initial medium lamination parameters and As the medium density in the elastic wave equation and , substitute into the elastic wave equation for calculation, and the initial synthetic seismic data can be obtained. The initial synthetic seismic data includes the horizontal component of the particle velocity vector and the vertical component of the particle velocity vector . Among them, the elastic wave equation includes:

[0059]

[0060]

[0061]

[0062]

[0063] Where, represents the density of the medium, Indicates the horizontal plane, represents the vertical surface, represents the horizontal component of the particle velocity vector, represents the seismic wave propagation time, express Stress response in the direction, express Stress response in the direction, represents the vertical component of the particle velocity vector, express Stress response in the direction, represents the medium Lamé parameter, Indicates the medium Lamé parameter.

[0064] In an optional implementation, step S150 may include the following steps: Step S151: Based on the longitudinal wave velocity, the shear wave velocity, the medium density, and the Lamé parameter calculation formula, the target medium Lamé parameter is obtained. The Lamé parameter calculation formula includes:

[0065]

[0066] Where, and Both represent the target medium Lamé parameters, represents the density of the medium, represents the shear wave velocity, represents the longitudinal wave velocity.

[0067] In step S151, the target medium Lamé parameter is calculated by using the longitudinal wave velocity, shear wave velocity, medium density and the Lamé parameter calculation formula. and .

[0068] The following examples of the present application further verify the application effect of the elastic wave full waveform inversion method of the present application through Example 1 for the complex structure elastic parameter model and Example 2 for the abnormal body elastic parameter model. The complex structure elastic parameter model and the abnormal body elastic parameter model are two typical problems in the elastic wave full waveform inversion.

[0069] Example 1: In order to verify the applicability of the elastic wave full waveform inversion method based on the single encoder-multiple decoder neural network structure proposed in this application in alleviating the dependence on the initial model, Example 1 uses a two-dimensional elastic Marmousi model to perform numerical simulation experiments. Figure 2 Schematically shows a schematic diagram of a real elastic parameter model based on a two-dimensional elastic Marmousi model simulation according to an embodiment of the present application, Figure 2 (a) is the longitudinal wave velocity model simulated based on the two-dimensional elastic Marmousi model. Figure 2 (b) is the shear wave velocity model simulated based on the two-dimensional elastic Marmousi model. Figure 2 (c) is the medium density model simulated based on the two-dimensional elastic Marmousi model. The model size of the two-dimensional elastic Marmousi model is 128×256, the spatial sampling interval is 20m, the shallow layer is set with a water layer of about 170m thick, and the longitudinal wave velocity is , the shear wave velocity is , the medium density is The observation system configuration consisted of 20 seismic sources evenly spaced along the upper boundary of the 2D elastic Marmousi model, and 256 geophones located at the bottom of the water layer. The source signal used a Ricker wavelet with a main frequency of 10 Hz, a time sampling interval of 6 ms, and a maximum recording time of 2.4 s.

[0070] Figure 3A schematic diagram of an initial elastic parameter model according to an embodiment of the present application is schematically shown. Figure 3 (d) is the initial longitudinal wave velocity model, Figure 3 (e) is the initial shear wave velocity model, Figure 3 (f) is the initial medium density model, and the first initial elastic parameter model is constructed by linear interpolation. A comparative experiment was conducted between the traditional physics-driven elastic wave full waveform inversion method and the elastic wave full waveform inversion method of the present application, and both used the Adam optimization algorithm for parameter update. The learning rate of the traditional physics-driven elastic wave full waveform inversion method was set to 5, and the learning rate of the elastic wave full waveform inversion method of the present application was set to 0.001. In order to further alleviate the crosstalk problem in the inversion process of multiple elastic parameters, the custom weights of the elastic wave full waveform inversion method of the present application are: the custom weight of the longitudinal wave velocity is 2, the custom weight of the shear wave velocity is 1, and the custom weight of the medium density is 0.04. The design of this custom weight can effectively control the update amplitude of the longitudinal wave velocity, shear wave velocity and medium density during the joint inversion process, thereby improving the stability and resolution of the parameter inversion.

[0071] In this embodiment, the Marmousi model is selected for comparative experiments. The same set of observed seismic data is input into the traditional physics-driven elastic wave full waveform inversion method and the multi-decoding neural network based on physical constraints constructed by the elastic wave full waveform inversion method of this application, and 2000 iterations are performed.

[0072] Figure 4 The schematic diagram of an elastic parameter model obtained based on the elastic wave full waveform inversion method of the present application according to an embodiment of the present application is shown schematically. Figure 4 As shown, Figure 4 (a) is a longitudinal wave velocity model obtained based on the elastic wave full waveform inversion method of this application, Figure 4 (b) is a shear wave velocity model obtained based on the elastic wave full waveform inversion method of this application, Figure 4 (c) is a medium density model obtained based on the elastic wave full waveform inversion method of this application.

[0073] Figure 5 The figure schematically shows an elastic parameter model obtained by inverting an elastic wave full waveform inversion method based on traditional physical drive according to an embodiment of the present application. Figure 5 (d) is a longitudinal wave velocity model obtained by the elastic wave full waveform inversion method based on traditional physical drive. Figure 5 (e) is a shear wave velocity model obtained by the elastic wave full waveform inversion method based on traditional physical drive. Figure 5 (f) A medium density model obtained by the elastic wave full waveform inversion method based on traditional physics drive.

[0074] Compare separately Figure 4 (a) to (c) and Figure 5 (e)-(f) Traditional physics-driven elastic wave full waveform inversion methods, due to their high reliance on the initial model, are prone to frequency skipping, resulting in inaccurate recovery of key model structures, particularly significant imaging deviations of shear wave velocity and medium density. In contrast, the elastic parameter model constructed using the elastic wave full waveform inversion method proposed in this application has a clear overall structure, no significant interference between elastic parameters, and exhibits greater adaptability and stability to the initial model.

[0075] To further evaluate the accuracy, the cross-section of each elastic parameter model was extracted at the 2560th common depth point (CDP=2.56km) at the underground horizontal position, and compared with the real elastic parameter model and the first initial elastic parameter model, and the corresponding single-channel curves of longitudinal wave velocity, shear wave velocity and medium density were drawn. Figure 6 Schematic diagram showing a comparison of single-channel elastic parameter curves according to an embodiment of the present application, as shown in FIG. Figure 6 As shown, Figure 6 (a) is the single-channel curve of longitudinal wave velocity, Figure 6 (b) is the single-channel curve of shear wave velocity, Figure 6 (c) is a single-channel mass density curve. It can be observed that the profile obtained by the elastic wave full waveform inversion method based on this application has a higher degree of fit with the true curve and has better performance in elastic parameter characterization than the traditional physical-driven elastic wave full waveform inversion method. At the same time, the change trend of the normalized loss function of the two methods is compared. Figure 7 A schematic diagram showing a comparison of the change trend of a normalized loss function according to an embodiment of the present application is shown as follows: Figure 7 As shown, although the traditional physics-driven elastic wave full waveform inversion method converges quickly in the initial stage, it tends to stagnate after iteration and there is a risk of converging in a non-optimal direction. The elastic wave full waveform inversion method of the present application maintains stable oscillation convergence, and the overall error reduction effect is more significant.

[0076] Example 2: In order to verify the applicability of the elastic wave full waveform inversion method based on the single encoder-multi-decoder neural network structure proposed in this application to the elastic parameter model of the abnormal body, the elastic wave full waveform inversion method of this application is applied to the two-dimensional elastic BP model with higher inversion difficulty and numerical simulation is carried out. Figure 8 Schematically shows a schematic diagram of a real elastic parameter model simulated based on a schematic diagram of a two-dimensional elastic BP model according to an embodiment of the present application, as shown in FIG. Figure 8 As shown, Figure 8 (a) is the longitudinal wave velocity model simulated based on the two-dimensional elastic BP model. Figure 8 (b) is the shear wave velocity model simulated based on the two-dimensional elastic BP model. Figure 8 (c) is a medium density model simulated based on a two-dimensional elastic BP model. The grid size of the BP model is 128×256, and the spatial sampling interval is 20m. The shallow part of the BP model contains a water layer with a thickness of about 170m. The elastic parameters of the BP model are: the longitudinal wave velocity is , the shear wave velocity is , the medium density is The observation system configuration consisted of 20 seismic sources evenly distributed across the top of the BP model and 256 geophones at the bottom of the water layer. The seismic source signal used was a Ricker wavelet with a primary frequency of 10 Hz, a time sampling interval of 6 ms, and a maximum recording duration of 3 s. Figure 9 Schematically shows another schematic diagram of an initial elastic parameter model according to an embodiment of the present application, Figure 9 (d) is the initial longitudinal wave velocity model, Figure 9 (e) is the initial shear wave velocity model, Figure 9 (f) is the initial medium density model, and the initial elastic parameter model is obtained by smoothing the real model.

[0077] Under the same optimization strategy, this embodiment compares and analyzes the traditional full waveform inversion method and the multi-decoding neural network method based on physical guidance proposed in the present invention, and both use the Adam algorithm for parameter iterative update. Among them, the learning rate of the traditional method is set to 5, while the method of the present invention uses a smaller learning rate of 0.001 to improve the stability of network training. In order to further alleviate the crosstalk problem in the inversion process of multiple elastic parameters, the custom weights of the elastic wave full waveform inversion method of the present application are: the custom weight of the longitudinal wave velocity is 2, the custom weight of the shear wave velocity is 1, and the custom weight of the medium density is 0.04. The design of this custom weight can effectively control the update amplitude of the longitudinal wave velocity, shear wave velocity and medium density in the joint inversion process, and improve the stability and resolution of the parameter inversion.

[0078] In this embodiment, the BP model is selected for comparative experiments, and the same set of observed seismic data is input into the multi-decoding neural network based on physical constraints constructed by the traditional physical-driven elastic wave full waveform inversion method and the elastic wave full waveform inversion method of this application, and 2000 iterations are performed.

[0079] Figure 10 Schematic diagram of another elastic parameter model obtained based on the elastic wave full waveform inversion method of the present application according to an embodiment of the present application is shown as follows: Figure 10 As shown, Figure 10 (a) is another longitudinal wave velocity model obtained based on the elastic wave full waveform inversion method of this application, Figure 10 (b) is another shear wave velocity model obtained based on the elastic wave full waveform inversion method of this application, Figure 10(c) is another medium density model obtained based on the elastic wave full waveform inversion method of this application.

[0080] Figure 11 Schematic diagram of an elastic parameter model obtained by inverting another elastic wave full waveform inversion method based on traditional physical drive according to an embodiment of the present application, Figure 11 (d) is another P-wave velocity model obtained by the elastic wave full waveform inversion method based on traditional physical drive. Figure 11 (e) is another shear wave velocity model obtained by the elastic wave full waveform inversion method based on traditional physical drive. Figure 11 (f) Another medium density model obtained by the elastic wave full waveform inversion method based on traditional physics drive.

[0081] The initial elastic parameter model selected in this embodiment is relatively ideal, and the traditional physics-driven elastic wave full waveform inversion method can achieve a certain degree of structural recovery in the salt dome area. However, the traditional physics-driven elastic wave full waveform inversion method is highly dependent on the initial elastic parameter model, and is prone to cycle skipping during the inversion process. In particular, there are significant errors in the reconstruction of shear wave velocity and medium density, which makes it difficult to accurately identify deep structures and key geological features. In contrast, the elastic wave full waveform inversion method based on a multi-decoder neural network of the present application shows a stronger modeling capability in complex subsalt structural areas. The inversion results obtained based on the elastic wave full waveform inversion method of the present application are highly consistent with the true model, especially in the characterization of medium density and deep structure. At the same time, there is no obvious crosstalk between multiple elastic parameters, indicating that the method has good parameter discrimination ability and stability.

[0082] Comparing the changing trends of the normalized loss functions of the two methods, Figure 12 A schematic diagram showing a comparison of another normalized loss function change trend according to an embodiment of the present application is shown as follows: Figure 12 As shown in the figure, due to the relatively ideal initial elastic parameter model, both methods showed a good convergence trend during the training process. However, further analysis showed that although the traditional physics-driven elastic wave full waveform inversion method decreased rapidly in the initial stage, the convergence speed of the traditional physics-driven elastic wave full waveform inversion method slowed down significantly in the later stage, and there was a risk of falling into local extreme values, making it difficult to continuously optimize the objective function. The elastic wave full waveform inversion method of the present application maintained a relatively stable oscillating convergence feature throughout the training process, and the overall error continued to decline, reflecting better global convergence ability and optimization stability.

[0083] The present application also provides an elastic wave full waveform inversion system, the system comprising: an acquisition module, configured to acquire initial synthetic seismic data, observed seismic data, and a first initial elastic parameter model, wherein the observed seismic data includes a horizontal component of an initial particle velocity vector and a vertical component of an initial particle velocity vector; An extraction module is used to input the observed seismic data into an encoder module of a preset neural network to extract a high-dimensional feature representation of the observed seismic data, wherein the preset neural network is trained by a Unet model; The first obtaining module is used to input the high-dimensional feature representation into the decoder module of the preset neural network to obtain the initial gradient update value of the elastic parameter; A second obtaining module is configured to obtain a first target elastic parameter model related to the longitudinal wave velocity, the shear wave velocity, and the medium density based on the initial gradient update value and the first initial elastic parameter model; A third obtaining module is used to obtain the target medium Lame parameters based on the longitudinal wave velocity, the shear wave velocity and the medium density; a fourth obtaining module, configured to obtain synthetic multi-component seismic data based on the target medium Lame parameter and the elastic wave equation, wherein the synthetic multi-component seismic data includes a horizontal component of the target particle velocity vector and a vertical component of the target particle velocity vector; a fifth obtaining module, for obtaining a target function based on the observed seismic data and the synthesized multi-component seismic data; an iterative update module for iteratively updating the weights of a preset neural network using an objective function until an error obtained based on the observed seismic data, the synthesized multi-component seismic data, and the loss function reaches a first preset threshold or the number of iterative updates reaches a second preset threshold, thereby obtaining a target gradient update value of the elastic parameter; The sixth obtaining module is used to obtain a final target elastic parameter model related to the target longitudinal wave velocity, target shear wave velocity and target medium density based on the target gradient update value and the second target elastic parameter model, wherein the second target elastic parameter model is determined based on the gradient update value and the elastic parameter model obtained by iteratively updating the weights of the previous preset neural network.

[0084] It is understandable that the elastic wave full waveform inversion system provided in the embodiment of the present application can implement each process of the elastic wave full waveform inversion method in the above embodiment and can achieve the same technical effect. To avoid repetition, it will not be described here.

[0085] The present application also provides an elastic wave full waveform inversion device, comprising: a memory configured to store instructions; The processor is configured to call instructions from the memory and implement the elastic wave full waveform inversion method as described above when executing the instructions.

[0086] It is understandable that the elastic wave full waveform inversion device provided in the embodiment of the present application can implement each process of the elastic wave full waveform inversion method in the above embodiment and can achieve the same technical effect. To avoid repetition, it will not be described here.

[0087] An embodiment of the present application further provides a machine-readable storage medium having stored thereon instructions for causing a machine to execute the elastic wave full waveform inversion method as described above.

[0088] It is understandable that the machine-readable storage medium provided in the embodiment of the present application can implement each process of the elastic wave full waveform inversion method in the above embodiment and can achieve the same technical effect. To avoid repetition, it will not be described here.

[0089] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0090] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0091] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0092] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0093] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0094] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0095] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can be implemented using any method or technology for information storage. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change RAM (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media, such as modulated data signals and carrier waves.

[0096] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0097] The above are merely embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A full waveform inversion method for elastic waves, characterized in that: The method comprises: Acquiring initial synthetic seismic data, observed seismic data, and a first initial elastic parameter model, wherein the observed seismic data includes a horizontal component of an initial particle velocity vector and a vertical component of an initial particle velocity vector; Inputting the observed seismic data into an encoder module of a preset neural network to extract a high-dimensional feature representation of the observed seismic data, wherein the preset neural network is trained by a Unet model; Inputting the high-dimensional feature representation into a decoder module of the preset neural network to obtain an initial gradient update value of the elasticity parameter; Obtaining a first target elastic parameter model related to longitudinal wave velocity, shear wave velocity, and medium density based on the initial gradient update value and the first initial elastic parameter model; Obtaining a target medium Lamé parameter based on the longitudinal wave velocity, the shear wave velocity, and the medium density; Based on the target medium Lame parameters and the elastic wave equation, synthetic multi-component seismic data is obtained, wherein the synthetic multi-component seismic data includes a horizontal component of a target particle velocity vector and a vertical component of a target particle velocity vector; obtaining an objective function based on the observed seismic data and the synthesized multi-component seismic data; Iteratively updating the weights of the preset neural network using the objective function until an error obtained based on the observed seismic data, the synthesized multi-component seismic data, and the loss function reaches a first preset threshold or the number of iterative updates reaches a second preset threshold, thereby obtaining a target gradient update value of the elastic parameter; Based on the target gradient update value and the second target elastic parameter model, a final target elastic parameter model related to the target longitudinal wave velocity, target shear wave velocity and target medium density is obtained, wherein the second target elastic parameter model is determined based on the gradient update value and elastic parameter model obtained by the previous iterative update of the weights of the preset neural network.

2. The method according to claim 1, characterized in that The first target elastic parameter model includes: Where, represents the first target elastic parameter model, represents the decoder module, represents the encoder module, represents the network weight, represents the observed seismic data, Indicates custom weights, represents the update step size, represents the first initial elastic parameter model.

3. The method according to claim 1, characterized in that The obtaining of target medium Lamé parameters based on the longitudinal wave velocity, the shear wave velocity and the medium density includes: The target medium Lamé parameter is obtained based on the longitudinal wave velocity, the shear wave velocity, the medium density, and a Lamé parameter calculation formula, wherein the Lamé parameter calculation formula includes: Where, and Both represent the target medium Lamé parameters, represents the medium density, represents the shear wave velocity, represents the longitudinal wave velocity.

4. The method according to claim 1, wherein The obtaining of initial synthetic seismic data comprises: Obtaining initial medium density, initial longitudinal wave velocity and initial shear wave velocity; Obtaining initial medium Lamé parameters based on the initial medium density, the initial longitudinal wave velocity, and the initial shear wave velocity; The initial synthetic seismic data is obtained based on the initial medium Lame parameters and the elastic wave equation.

5. The method according to claim 4, characterized in that The elastic wave equation includes: Where, represents the density of the medium, Indicates the horizontal plane, represents the vertical surface, represents the horizontal component of the particle velocity vector, represents the seismic wave propagation time, express Stress response in the direction, express Stress response in the direction, represents the vertical component of the particle velocity vector, express Stress response in the direction, represents the medium Lamé parameter, Indicates the medium Lamé parameter.

6. The method according to claim 1, characterized in that The encoder module includes an encoder, which is composed of multiple 3*3 convolutional layers and 2*2 maximum pooling layers alternating.

7. The method according to claim 1, characterized in that The decoder module includes a first decoder, a second decoder and a third decoder. The first decoder is used to generate the longitudinal wave velocity, the second decoder is used to generate the shear wave velocity, and the third decoder is used to generate the medium density. The first decoder, the second decoder and the third decoder are all composed of multiple 2*2 deconvolution layers.

8. An elastic wave full waveform inversion system, characterized in that: The system comprises: an acquisition module, configured to acquire initial synthetic seismic data, observed seismic data, and a first initial elastic parameter model, wherein the observed seismic data includes a horizontal component of an initial particle velocity vector and a vertical component of an initial particle velocity vector; an extraction module, configured to input the observed seismic data into an encoder module of a preset neural network to extract a high-dimensional feature representation of the observed seismic data, wherein the preset neural network is trained by a Unet model; A first obtaining module is configured to input the high-dimensional feature representation into a decoder module of the preset neural network to obtain an initial gradient update value of the elasticity parameter; a second obtaining module, configured to obtain a first target elastic parameter model related to longitudinal wave velocity, shear wave velocity, and medium density based on the initial gradient update value and the first initial elastic parameter model; A third obtaining module is used to obtain the target medium Lamé parameter based on the longitudinal wave velocity, the shear wave velocity and the medium density; a fourth obtaining module, configured to obtain synthetic multi-component seismic data based on the target medium Lame parameter and the elastic wave equation, wherein the synthetic multi-component seismic data includes a horizontal component of a target particle velocity vector and a vertical component of a target particle velocity vector; a fifth obtaining module, configured to obtain an objective function based on the observed seismic data and the synthesized multi-component seismic data; an iterative update module, configured to iteratively update the weights of the preset neural network using the objective function until an error obtained based on the observed seismic data, the synthesized multi-component seismic data, and the loss function reaches a first preset threshold or the number of iterative updates reaches a second preset threshold, thereby obtaining a target gradient update value of the elastic parameter; The sixth obtaining module is used to obtain a final target elastic parameter model related to the target longitudinal wave velocity, target shear wave velocity and target medium density based on the target gradient update value and the second target elastic parameter model, wherein the second target elastic parameter model is determined based on the gradient update value and the elastic parameter model obtained by the previous iterative update of the weights of the preset neural network.

9. An elastic wave full waveform inversion device, characterized in that: include: a memory configured to store instructions; A processor is configured to call the instructions from the memory and implement the elastic wave full waveform inversion method according to any one of claims 1 to 7 when executing the instructions.

10. A machine-readable storage medium, characterized in that The machine-readable storage medium stores instructions for causing a machine to execute the elastic wave full waveform inversion method according to any one of claims 1 to 7.

Images