Tomographic inversion method, apparatus, electronic device, and storage medium

Abstract

The application provides a tomographic inversion method and device, electronic equipment and storage medium. The method comprises: obtaining shot point spatial coordinates, multiple receiver point spatial coordinates and shot-receiver point travel time data; creating a partial differential equation solving space according to the shot point spatial coordinates and the receiver point spatial coordinates, and discretizing to obtain partial differential equation space coordinates; determining a travel time data error of the shot-receiver point travel time data under a constructed deep learning neural network, and a partial differential equation error in the partial differential equation solving space; using an augmented Lagrangian framework, fusing the travel time data error and the partial differential equation error, and constructing a target velocity field prediction neural network; inputting the partial differential equation space coordinates into the target velocity field prediction neural network to obtain a tomographic velocity model. Thus, the travel time field and the velocity field are simultaneously predicted and constrained, and the velocity model is updated.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of seismic exploration data processing, and in particular to a tomographic inversion method and device, an electronic device and a storage medium. BACKGROUND

[0002] Shallow velocity modeling is an important step in the industry's seismic exploration data processing flow, and tomographic inversion can not only provide a velocity model for static correction calculation, but also provide an initial velocity model for reverse time migration (RTM) and full waveform inversion.

[0003] At the present stage, conventional tomographic inversion is usually based on a travel time forward operator under the condition of equal-interval fixed grid or variable grid. In the conventional tomographic inversion process, an initial velocity model is first given, and then the predicted travel time under the velocity model is calculated using the travel time forward operator. Next, the predicted travel time is compared with the observed travel time, and the velocity model is iteratively adjusted according to the gradient information by calculating the gradient of the travel time, so that the predicted travel time gradually approaches the observed travel time. The iteration process will continue until a certain iteration convergence criterion is met, and finally a velocity model that fits the observed data is obtained.

[0004] With the rise of neural networks, there are also data-driven design-based velocity modeling techniques that calculate corresponding travel time curves through conventional function equation algorithms, generate various velocity models, and then use the velocity models and travel time curves as sample labels for neural network training. This data-driven approach not only has high requirements for the training set, but also is not flexible enough for the observation system, that is, when the observation system changes, it is uncertain whether the neural network can infer a reasonable result. SUMMARY

[0005] The present application provides a tomographic inversion method, device, electronic device and storage medium for shortening the time required for tomographic inversion while ensuring the accuracy of the underground medium velocity model.

[0006] In a first aspect, the present application provides a tomographic inversion method, comprising:

[0007] obtaining shot point spatial coordinates, a plurality of receiver point spatial coordinates and shot-receiver point travel time data;

[0008] creating a partial differential equation solving space according to the shot point spatial coordinates and the receiver point spatial coordinates, and discretely obtaining partial differential equation space coordinates;

[0009] determining the travel time data error of the shot-receiver point travel time data under the constructed deep learning neural network, and the partial differential equation error in the partial differential equation solving space;

[0010] In the augmented Lagrangian framework, the travel time data error and the partial differential equation error are fused to construct a target velocity field prediction neural network.

[0011] The partial differential equation spatial coordinates are input into the target velocity field prediction neural network to obtain a tomographic velocity model.

[0012] Optionally, in the augmented Lagrangian framework, the travel time data error and the partial differential equation error are fused to construct a target velocity field prediction neural network, comprising:

[0013] In combination with the travel time data error, the partial differential equation error, and the boundary condition of the partial differential equation model, an augmented Lagrangian function under the augmented Lagrangian framework is constructed.

[0014] Taking the augmented Lagrangian function as a target function, an optimization algorithm is used to update network weights in the training process of the deep learning neural network to obtain the target velocity field prediction neural network.

[0015] Optionally, the deep learning neural network is an independent double-network structure, comprising a travel time field prediction neural network and a velocity field prediction neural network; the travel time data error of the shot-receiver point travel time data under the constructed deep learning neural network and the partial differential equation error in the partial differential equation solving space are determined, comprising:

[0016] The shot point spatial coordinates, the receiver point spatial coordinates, and the shot-receiver point travel time data are input into the travel time field prediction neural network to obtain predicted travel time data.

[0017] The partial differential equation spatial coordinates are input into the velocity field prediction neural network to obtain a predicted velocity field.

[0018] Based on the predicted travel time data and the predicted velocity field, the travel time data error and the partial differential equation error are determined.

[0019] Optionally, based on the predicted travel time data and the predicted velocity field, the travel time data error and the partial differential equation error are determined, comprising:

[0020] The travel time field error of each shot-receiver point in the predicted travel time data and the corresponding shot-receiver point travel time data in the partial differential equation solving space is calculated.

[0021] Based on a pre-set partial differential equation error calculation expression, in combination with the predicted velocity field and the predicted travel time data, the partial differential equation error is obtained.

[0022] In a second aspect, the present application provides a tomographic inversion device, comprising:

[0023] an acquisition module configured to acquire shot point spatial coordinates, a plurality of receiver point spatial coordinates, and shot-receiver point travel time data;

[0024] a solution space creation module configured to create a partial differential equation solution space according to the shot point spatial coordinates and the receiver point spatial coordinates, and discretize to obtain partial differential equation spatial coordinates;

[0025] an error determination module configured to determine a travel time data error of the shot-receiver point travel time data under a constructed deep learning neural network, and a partial differential equation error of the partial differential equation solution space;

[0026] a network construction module configured to fuse the travel time data error and the partial differential equation error in an augmented Lagrangian framework to construct a target velocity field prediction neural network;

[0027] a tomography module configured to input the partial differential equation spatial coordinates into the target velocity field prediction neural network to obtain a tomographic velocity model.

[0028] Optionally, the network construction module comprises:

[0029] a function construction submodule configured to construct an augmented Lagrangian function under the augmented Lagrangian framework in combination with the travel time data error, the partial differential equation error, and a boundary condition of a partial differential equation model;

[0030] a network construction submodule configured to use an optimization algorithm to update network weights in a training process of the deep learning neural network with the augmented Lagrangian function as a target function, to obtain the target velocity field prediction neural network.

[0031] Optionally, the deep learning neural network is an independent double-network structure, comprising a travel time field prediction neural network and a velocity field prediction neural network; and the error determination module comprises:

[0032] a travel time data prediction submodule configured to input the shot point spatial coordinates, the receiver point spatial coordinates, and the shot-receiver point travel time data into the travel time field prediction neural network to obtain predicted travel time data;

[0033] a velocity field prediction submodule configured to input the partial differential equation spatial coordinates into the velocity field prediction neural network to obtain a predicted velocity field;

[0034] an error determination submodule configured to determine the travel time data error and the partial differential equation error based on the predicted travel time data and the predicted velocity field.

[0035] Optionally, the error determination submodule comprises:

[0036] a walk time field error calculation unit configured to calculate, in the partial differential equation solving space, a walk time field error of each shot-receiver point under the predicted walk time data and corresponding shot-receiver point walk time data;

[0037] a partial differential equation error calculation unit configured to obtain the partial differential equation error based on a pre-set partial differential equation error calculation expression, in combination with the predicted velocity field and the predicted walk time data.

[0038] In a third aspect, the present application provides an electronic device, comprising a processor and a memory, wherein the memory stores computer readable instructions, and when the computer readable instructions are executed by the processor, the steps in the method provided in the first aspect above are executed.

[0039] In a fourth aspect, the present application provides a storage medium, which stores a computer program, and when the computer program is executed by a processor, the steps in the method provided in the first aspect above are executed.

[0040] From the above technical solutions, the present application has the following advantages:

[0041] The present application provides a tomographic inversion method and device, an electronic device and a storage medium. The method comprises: obtaining shot point spatial coordinates, multiple receiver spatial coordinates and shot-receiver point walk time data; creating a partial differential equation solving space according to the shot point spatial coordinates and the receiver spatial coordinates, and discretely obtaining partial differential equation spatial coordinates; determining a walk time data error of the shot-receiver point walk time data under a deep learning neural network and a partial differential equation error in the partial differential equation solving space; using an augmented Lagrangian framework to fuse the walk time data error and the partial differential equation error to construct a target velocity field prediction neural network; inputting the partial differential equation spatial coordinates into the target velocity field prediction neural network to obtain a tomographic velocity model. The target velocity field prediction neural network is constructed by using the augmented Lagrangian framework, the data error and the partial differential equation error fusion, the simultaneous prediction and constraint of the walk time field and the velocity field are realized, and the velocity model is updated. BRIEF DESCRIPTION OF DRAWINGS

[0042] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0043] Figure 1 Flow chart of a tomographic inversion method embodiment one of the present application;

[0044] Figure 2 Figure 1 is a flow chart of an embodiment of a tomographic inversion method of the present application;

[0045] Figure 3 Figure 2 is a flow chart of an embodiment of a tomographic inversion method of the present application;

[0046] Figure 4 Figure 3 is a model topography diagram of an embodiment of a tomographic inversion method of the present application;

[0047] Figure 5 Figure 4 is a real model longitudinal profile diagram of an embodiment of a tomographic inversion method of the present application;

[0048] Figure 6 Figure 5 is a tomographic velocity diagram of an embodiment of a tomographic inversion method of the present application;

[0049] Figure 7 Figure 6 is a tomographic velocity versus real velocity comparison diagram of an embodiment of a tomographic inversion method of the present application;

[0050] Figure 8 Figure 7 is a structural block diagram of an embodiment of a tomographic inversion device of the present application. DETAILED DESCRIPTION

[0051] The embodiment of the present application provides a tomographic inversion method, device, electronic equipment and storage medium, which is used for shortening the time required for tomographic inversion while ensuring the accuracy of the underground medium velocity model.

[0052] In order to make the purpose, features and advantages of the present application more obvious and easy to understand, 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. Obviously, the embodiments described below are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0053] Embodiment 1, please refer to Figure 1 , Figure 1 Figure 1 is a flow chart of an embodiment of a tomographic inversion method of the present application, comprising:

[0054] Step S101, obtaining a shot point spatial coordinate, a plurality of receiver point spatial coordinates and shot-receiver point travel time data;

[0055] Step S102, creating a partial differential equation solving space according to the shot point spatial coordinate and the receiver point spatial coordinate, and discretizing to obtain a partial differential equation spatial coordinate;

[0056] Step S103, determining the travel time data error of the shot-receiver point travel time data under the constructed deep learning neural network and the partial differential equation error in the partial differential equation solving space;

[0057] Step S104, constructing a target velocity field prediction neural network by fusing the travel time data error and the partial differential equation error in an augmented Lagrange framework;

[0058] Step S105, inputting the partial differential equation space coordinates into the target velocity field prediction neural network to obtain a tomographic velocity model.

[0059] In the tomographic inversion method provided in the embodiment of the application, the shot point space coordinates, the multiple receiver point space coordinates and the shot-receiver point travel time data are obtained; the partial differential equation solving space is created according to the shot point space coordinates and the receiver point space coordinates, and the partial differential equation space coordinates are obtained by discretization; the travel time data error of the shot-receiver point travel time data under the constructed deep learning neural network and the partial differential equation error in the partial differential equation solving space are determined; the target velocity field prediction neural network is constructed by fusing the travel time data error and the partial differential equation error in an augmented Lagrange framework; and the partial differential equation space coordinates are input into the target velocity field prediction neural network to obtain a tomographic velocity model. The target velocity field prediction neural network is constructed by fusing the augmented Lagrange framework, the data error and the partial differential equation error, the travel time field and the velocity field are simultaneously predicted and constrained, and the velocity model is updated.

[0060] Embodiment two, please refer to Figure 2 , Figure 2 is a flow step diagram of embodiment two of the tomographic inversion method of the application, comprising:

[0061] Step S201, obtaining shot point space coordinates, multiple receiver point space coordinates and shot-receiver point travel time data;

[0062] Step S202, creating a partial differential equation solving space according to the shot point space coordinates and the receiver point space coordinates, and obtaining partial differential equation space coordinates by discretization;

[0063] Step S203, determining the travel time data error of the shot-receiver point travel time data under the constructed deep learning neural network and the partial differential equation error in the partial differential equation solving space;

[0064] Step S204, constructing an augmented Lagrange function under the augmented Lagrange framework by combining the travel time data error, the partial differential equation error and the boundary conditions of the partial differential equation model;

[0065] Step S205, using an optimization algorithm, updating the network weight in the training process of the deep learning neural network with the augmented Lagrange function as the objective function, to obtain the target velocity field prediction neural network;

[0066] Step S206, inputting the partial differential equation space coordinates into the target velocity field prediction neural network to obtain the tomographic velocity model.

[0067] In the tomographic inversion method provided in the embodiment of the application, by introducing a Lagrange multiplier and an augmented term, combining data errors, partial differential equation errors, and boundary conditions of the partial differential equation model, an augmented Lagrange function is constructed, and the prediction neural network is trained with the augmented Lagrange function as an objective function, to obtain a target velocity field prediction neural network that simultaneously satisfies data constraints and physical constraints and can provide accurate tomographic inversion.

[0068] Embodiment three, please refer to Figure 3 , Figure 3 is a flow step diagram of an embodiment four of the tomographic inversion method of the application, comprising:

[0069] Step S301, acquiring shot point space coordinates, multiple receiver point space coordinates, and shot-receiver point travel time data;

[0070] Step S302, creating a partial differential equation solving space according to the shot point space coordinates and the receiver point space coordinates, and discretizing to obtain partial differential equation space coordinates;

[0071] Step S303, inputting the shot point space coordinates, the receiver point space coordinates, and the shot-receiver point travel time data into the travel time field prediction neural network to obtain predicted travel time data;

[0072] Step S304, inputting the partial differential equation space coordinates into the velocity field prediction neural network to obtain a predicted velocity field;

[0073] In the embodiment of the application, the prediction neural network adopts an independent double-network structure to fit the travel time field and the velocity field respectively, that is:

[0074]

[0075] wherein, is a space point coordinate, and N t is the travel time field prediction neural network, which can be understood as a composite function; N vThe velocity field prediction neural network can also be understood as a composite function. The corresponding physical meaning is that for any spatial point, the travel time field prediction neural network can give a travel time data from the source to the current spatial point, and the velocity field prediction neural network can give the velocity value of the current spatial point. In summary, the intelligent tomographic inversion problem is to obtain the velocity field prediction neural network and the travel time field prediction neural network simultaneously after the neural network training.

[0076] Step S305, calculating the travel time field error of each shot point in the partial differential equation solving space under the predicted travel time data and the corresponding shot point travel time data;

[0077] Step S306, obtaining the partial differential equation error based on the pre-set partial differential equation error calculation expression, combining the predicted velocity field and the predicted travel time data;

[0078] Step S307, combining the travel time data error, the partial differential equation error, and the boundary condition of the partial differential equation model, constructing the augmented Lagrangian function under the augmented Lagrangian framework;

[0079] The embodiment of the application adopts the augmented Lagrangian framework, and the data error and the partial differential equation error are fused together. For a single source, the corresponding augmented Lagrangian function is:

[0080]

[0081] Wherein, is the data error, is the partial differential equation error,

[0082] is the boundary condition of the partial differential equation. Specifically, N r is the number of receivers, is the value of the travel time field at the receiver position, is the observed travel time, N is the number of points of the discrete velocity field, T is the travel time field, v is the velocity field, and if the travel time is at the source position, is zero.

[0083] Step S308, taking the augmented Lagrangian function as the objective function, using an optimization algorithm to update the network weight in the training process of the deep learning neural network, and obtaining the target velocity field prediction neural network;

[0084] In the embodiment of the present application, the travel time curve is approximately positively correlated with the offset distance, and when the receiver coincides with the shot point, the travel time is zero; the farther the receiver is from the shot point, the greater the travel time. In the process of solving the intelligent tomography, the model may not be accurate enough at large offsets due to the unbalanced travel time data. Therefore, in order to accelerate the convergence speed of the intelligent tomography inversion, an analytical background travel time function can be given in advance, such as a function proportional to the offset distance, and in the training of the prediction neural network, the travel time field prediction neural network is pre-decomposed, the decomposition time variable can adopt the product T=T0τ or the addition T=T0+τ, and a travel time field prediction network about the intermediate variable τ is constructed to replace the travel time field prediction network about T. Thus, the value range of the intermediate variable becomes smaller, for example, for the travel time data generated by the uniform model, the travel time data is proportional to the distance; after decomposition by the product method, the true value of the intermediate variable is 1, and the contribution of each offset to the tomography inversion is balanced.

[0085] At the same time, in order to accelerate the training speed of the intelligent tomography inversion, for a single shot gather, the inversion range is automatically selected according to the coverage range of the receiver, so as to avoid inputting the entire work area coordinates, thereby reducing the demand for memory while accelerating the network training.

[0086] Further, the seismic exploration work area usually contains thousands of shot gathers, and it is difficult to read all the shot gathers at one time under the condition of limited computing resources, at this time, the incremental data processing method is needed to solve the mass data problem, that is, the data is input into the neural network in batches, and each neural network training is updated on the previous network model.

[0087] In addition, the embodiment of the present application adopts data component progressive application to solve the problem of slow neural network training in the case that the function relationship of the travel time curve, i.e. the first arrival travel time, about the offset distance or the offset distance is complex, that is, the observed data is first fitted by a polynomial, and the neural network is trained on the fitted data; after the training is completed, the high-frequency data component is gradually added until the original first arrival travel time is added. Thus, in the case that the geological model is complex and the travel time curve is complex and variable, if the travel time curve is directly input into the prediction neural network, the high-frequency component of the travel time curve needs to be iterated for multiple rounds before being fitted.

[0088] Finally, in order to reduce unnecessary calculation, the embodiment of the present application allows the user to specify the lower interface.

[0089] Step S309: inputting the partial differential equation space coordinates into the target velocity field prediction neural network to obtain a tomographic velocity model.

[0090] A tomographic inversion method, by acquiring shot point spatial coordinates, multiple receiver spatial coordinates and shot-receiver travel time data; according to the shot point spatial coordinates and the receiver spatial coordinates, a partial differential equation solving space is created, and the partial differential equation space coordinates are obtained by discretization; the travel time data error of the shot-receiver travel time data under the constructed deep learning neural network, and the partial differential equation error in the partial differential equation solving space are determined; using the augmented Lagrangian framework, the travel time data error and the partial differential equation error are fused to construct a target velocity field prediction neural network; the partial differential equation space coordinates are input into the target velocity field prediction neural network to obtain a tomographic velocity model. Through the augmented Lagrangian framework, the data error and the partial differential equation error fusion to construct the target velocity field prediction neural network, the simultaneous prediction and constraint of the travel time field and the velocity field are realized, and the velocity model is updated.

[0091] In order to facilitate the understanding of the beneficial effects of the embodiments of the present application by those skilled in the art, the velocity model obtained by the tomographic inversion of the fourth embodiment of the present application is compared with the real model and illustrated by examples.

[0092] Please refer to Figures 4-5 , Figure 4 The model topography diagram of the fourth embodiment of the tomographic inversion method of the present application is shown in Figure 5 The real model longitudinal profile diagram of the fourth embodiment of the tomographic inversion method of the present application is shown in Figure 4 And Figure 5 The target travel time field prediction neural network and the target velocity field prediction neural network are used for prediction to obtain the tomographic velocity diagram shown in Figure 6 Then, through the interval display comparison method, the tomographic velocity and real velocity comparison diagram of Figure 7 It can be seen that the velocity model obtained by the tomographic inversion method of the embodiment of the present application is consistent with the real velocity model.

[0093] Embodiment four, please refer to Figure 8 , Figure 8 The structure block diagram of the tomographic inversion device embodiment of the present application is shown in

[0094] The acquisition module 501 is used for acquiring shot point spatial coordinates, multiple receiver spatial coordinates and shot-receiver travel time data;

[0095] The solving space creation module 502 is used for creating a partial differential equation solving space according to the shot point spatial coordinates and the receiver spatial coordinates, and obtaining partial differential equation space coordinates by discretization;

[0096] The error determination module 503 is configured to determine a travel time data error of the shot-receiver point travel time data under the constructed deep learning neural network and a partial differential equation error in the partial differential equation solving space;

[0097] The network construction module 504 is configured to construct a target velocity field prediction neural network by fusing the travel time data error and the partial differential equation error in an augmented Lagrangian framework;

[0098] The tomography module 505 is configured to input the partial differential equation space coordinates into the target velocity field prediction neural network to obtain a tomographic velocity model.

[0099] In an optional embodiment, the network construction module 504 comprises:

[0100] The function construction submodule is configured to construct an augmented Lagrangian function in the augmented Lagrangian framework by combining the travel time data error, the partial differential equation error, and a boundary condition of a partial differential equation model;

[0101] The network construction submodule is configured to update network weights in a training process of the deep learning neural network by using an optimization algorithm with the augmented Lagrangian function as a target function, to obtain the target velocity field prediction neural network.

[0102] In an optional embodiment, the deep learning neural network is an independent double-network structure, comprising a travel time field prediction neural network and a velocity field prediction neural network; and the error determination module 503 comprises:

[0103] The travel time data prediction submodule is configured to input the shot point space coordinates, the receiver point space coordinates, and the shot-receiver point travel time data into the travel time field prediction neural network to obtain predicted travel time data;

[0104] The velocity field prediction submodule is configured to input the partial differential equation space coordinates into the velocity field prediction neural network to obtain a predicted velocity field;

[0105] The error determination submodule is configured to determine the travel time data error and the partial differential equation error based on the predicted travel time data and the predicted velocity field.

[0106] In an optional embodiment, the error determination submodule comprises:

[0107] The travel time field error calculation unit is configured to calculate, in the partial differential equation solving space, a travel time field error of each shot-receiver point under the predicted travel time data and corresponding shot-receiver point travel time data;

[0108] A partial differential equation error calculation unit is configured to obtain the partial differential equation error based on a preset partial differential equation error calculation expression, in combination with the predicted velocity field and the predicted travel time data.

[0109] In an embodiment, the present application also provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer program is executed by the processor to make the processor execute the steps of the tomographic inversion method, comprising:

[0110] obtaining spatial coordinates of a shot point, spatial coordinates of a plurality of receiver points, and travel time data of shot-receiver points;

[0111] creating a partial differential equation solving space according to the spatial coordinates of the shot point and the spatial coordinates of the receiver points, and discretely obtaining partial differential equation spatial coordinates;

[0112] determining travel time data errors of the travel time data of the shot-receiver points under the constructed deep learning neural network, and partial differential equation errors in the partial differential equation solving space;

[0113] using an augmented Lagrangian framework to fuse the travel time data errors and the partial differential equation errors to construct a target velocity field prediction neural network;

[0114] inputting the partial differential equation spatial coordinates into the target velocity field prediction neural network to obtain a tomographic velocity model.

[0115] In an optional embodiment, using an augmented Lagrangian framework to fuse the travel time data errors and the partial differential equation errors to construct a target velocity field prediction neural network, comprises:

[0116] constructing an augmented Lagrangian function under the augmented Lagrangian framework in combination with the travel time data errors, the partial differential equation errors, and boundary conditions of a partial differential equation model;

[0117] using an optimization algorithm to update network weights in a training process of the deep learning neural network to obtain the target velocity field prediction neural network, with the augmented Lagrangian function as an objective function.

[0118] In an optional embodiment, the deep learning neural network is an independent double-network structure, comprising a travel time field prediction neural network and a velocity field prediction neural network; and determining the travel time data errors of the travel time data of the shot-receiver points under the constructed deep learning neural network, and the partial differential equation errors in the partial differential equation solving space, comprises:

[0119] inputting the spatial coordinates of the shot point and the spatial coordinates of the receiver points and the travel time data of the shot-receiver points into the travel time field prediction neural network to obtain predicted travel time data;

[0120] inputting the partial differential equation space coordinates into the velocity field prediction neural network to obtain a predicted velocity field;

[0121] determining the travel time data error and the partial differential equation error based on the predicted travel time data and the predicted velocity field.

[0122] In an optional embodiment, determining the travel time data error and the partial differential equation error based on the predicted travel time data and the predicted velocity field comprises:

[0123] calculating a travel time field error of each shot-receiver point in the partial differential equation solving space under the predicted travel time data and corresponding shot-receiver point travel time data;

[0124] obtaining the partial differential equation error based on a preset partial differential equation error calculation expression in combination with the predicted velocity field and the predicted travel time data.

[0125] Embodiment six, the embodiment of the application further provides a computer storage medium, which has a computer program stored thereon, and the computer program is executed by the processor to realize steps of a tomographic inversion method, comprising:

[0126] obtaining shot point space coordinates, multiple receiver point space coordinates and shot-receiver point travel time data;

[0127] creating a partial differential equation solving space according to the shot point space coordinates and the receiver point space coordinates, and discretely obtaining partial differential equation space coordinates;

[0128] determining a travel time data error of the shot-receiver point travel time data under a constructed deep learning neural network and a partial differential equation error in the partial differential equation solving space;

[0129] constructing a target velocity field prediction neural network by fusing the travel time data error and the partial differential equation error in an augmented Lagrangian framework;

[0130] inputting the partial differential equation space coordinates into the target velocity field prediction neural network to obtain a tomographic velocity model.

[0131] In an optional embodiment, constructing a target velocity field prediction neural network by fusing the travel time data error and the partial differential equation error in an augmented Lagrangian framework comprises:

[0132] constructing an augmented Lagrangian function under the augmented Lagrangian framework in combination with the travel time data error, the partial differential equation error and boundary conditions of a partial differential equation model;

[0133] Taking the augmented Lagrange function as a target function, network weights are updated in a training process of the deep learning neural network by using an optimization algorithm, to obtain the target velocity field prediction neural network.

[0134] In an optional embodiment, the deep learning neural network is an independent double-network structure, comprising a travel time field prediction neural network and a velocity field prediction neural network; the travel time data error of the shot-receiver point travel time data under the constructed deep learning neural network and the partial differential equation error in the partial differential equation solving space are determined, comprising:

[0135] The shot point space coordinates, the receiver point space coordinates and the shot-receiver point travel time data are input into the travel time field prediction neural network to obtain predicted travel time data;

[0136] The partial differential equation space coordinates are input into the velocity field prediction neural network to obtain a predicted velocity field;

[0137] Based on the predicted travel time data and the predicted velocity field, the travel time data error and the partial differential equation error are determined.

[0138] In an optional embodiment, based on the predicted travel time data and the predicted velocity field, the travel time data error and the partial differential equation error are determined, comprising:

[0139] The travel time field error of each shot-receiver point under the predicted travel time data and the corresponding shot-receiver point travel time data in the partial differential equation solving space is calculated;

[0140] Based on a pre-set partial differential equation error calculation expression, the predicted velocity field and the predicted travel time data are combined to obtain the partial differential equation error.

[0141] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, device and unit can refer to the corresponding process in the foregoing method embodiments, which will not be repeated here.

[0142] In several embodiments provided in the present application, it should be understood that the disclosed methods, devices, electronic devices and storage media can be implemented by other ways. For example, the above-described device embodiments are only illustrative, for example, the division of the units is only a logical function division, and actual implementation can have another division way, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the displayed or discussed units can be indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0143] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purposes of the embodiments of the present application.

[0144] In addition, the functional units in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0145] The integrated unit, if realized in the form of a software functional unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the essential part or all or part of the technical solutions that contribute to the prior art can be embodied in the form of a software product. The computer software product is stored in a readable storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned readable storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various program code storage media.

[0146] The above-described embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or make equivalent replacements for some technical features; and these modifications or replacements do not make the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A tomographic inversion method characterized by, The method comprises the following steps: obtaining spatial coordinates of a shot point, spatial coordinates of multiple receiver points, and travel time data of shot-receiver points; creating a partial differential equation solving space according to the spatial coordinates of the shot point and the spatial coordinates of the receiver points, and discretizing to obtain partial differential equation space coordinates; determining travel time data errors of the travel time data of the shot-receiver points under the constructed deep learning neural network, and partial differential equation errors in the partial differential equation solving space; using an augmented Lagrangian framework to fuse the travel time data errors and the partial differential equation errors to construct a target velocity field prediction neural network; inputting the partial differential equation space coordinates into the target velocity field prediction neural network to obtain a tomographic velocity model; using an augmented Lagrangian framework to fuse the travel time data errors and the partial differential equation errors to construct a target velocity field prediction neural network, comprising: combining the travel time data errors, the partial differential equation errors, and boundary conditions of a partial differential equation model to construct an augmented Lagrangian function under the augmented Lagrangian framework; using an optimization algorithm to update network weights in a training process of the deep learning neural network with the augmented Lagrangian function as a target function, so as to obtain the target velocity field prediction neural network.

2. The tomographic inversion method of claim 1, wherein, The deep learning neural network is an independent double-network structure, comprising a travel time field prediction neural network and a velocity field prediction neural network; determining travel time data errors of the travel time data of the shot-receiver points under the constructed deep learning neural network, and partial differential equation errors in the partial differential equation solving space, comprising: inputting the spatial coordinates of the shot point, the spatial coordinates of the multiple receiver points, and the travel time data of the shot-receiver points into the travel time field prediction neural network to obtain predicted travel time data; inputting the partial differential equation space coordinates into the velocity field prediction neural network to obtain a predicted velocity field; determining the travel time data errors and the partial differential equation errors based on the predicted travel time data and the predicted velocity field.

3. The tomographic inversion method of claim 2, wherein, determining the travel time data errors and the partial differential equation errors based on the predicted travel time data and the predicted velocity field, comprising: calculating, in the partial differential equation solving space, travel time field errors of each shot-receiver point under the predicted travel time data and corresponding travel time data of the shot-receiver point; obtaining the partial differential equation errors based on a pre-set partial differential equation error calculation expression, in combination with the predicted velocity field and the predicted travel time data.

4. A tomographic inversion apparatus characterized by comprising: The method comprises the following steps: an obtaining module, configured to obtain spatial coordinates of a shot point, spatial coordinates of multiple receiver points, and travel time data of shot-receiver points; a solving space creating module, configured to create a partial differential equation solving space according to the spatial coordinates of the shot point and the spatial coordinates of the receiver points, and discretize to obtain partial differential equation space coordinates; an error determining module, configured to determine travel time data errors of the travel time data of the shot-receiver points under the constructed deep learning neural network, and partial differential equation errors in the partial differential equation solving space; a network constructing module, configured to use an augmented Lagrangian framework to fuse the travel time data errors and the partial differential equation errors to construct a target velocity field prediction neural network; a tomography module configured to input the partial differential equation spatial coordinates into the target velocity field prediction neural network to obtain a tomography velocity model; the network construction module comprises: a function construction submodule configured to construct an augmented Lagrangian function under the augmented Lagrangian framework by combining the travel time data error, the partial differential equation error, and a boundary condition of a partial differential equation model; a network construction submodule configured to use an optimization algorithm to update network weights in a training process of the deep learning neural network by taking the augmented Lagrangian function as a target function, to obtain the target velocity field prediction neural network.

5. The tomographic inversion apparatus according to claim 4, characterized by the deep learning neural network is an independent double-network structure, comprising a travel time field prediction neural network and a velocity field prediction neural network; the error determination module comprises: a travel time data prediction submodule configured to input the shot point spatial coordinates, the receiver point spatial coordinates, and the shot-receiver travel time data into the travel time field prediction neural network to obtain predicted travel time data; a velocity field prediction submodule configured to input the partial differential equation spatial coordinates into the velocity field prediction neural network to obtain a predicted velocity field; an error determination submodule configured to determine the travel time data error and the partial differential equation error based on the predicted travel time data and the predicted velocity field.

6. The tomographic inversion apparatus according to claim 5, wherein the error determination submodule comprises: a travel time field error calculation unit configured to calculate, in the partial differential equation solving space, a travel time field error of each shot-receiver point under the predicted travel time data and corresponding shot-receiver travel time data; a partial differential equation error calculation unit configured to obtain the partial differential equation error by combining the predicted velocity field and the predicted travel time data based on a pre-set partial differential equation error calculation expression.

7. An electronic device, comprising: a processor and a memory, the memory storing computer readable instructions, when the computer readable instructions are executed by the processor, the method of any one of claims 1-3 is run.

8. A storage medium having stored thereon a computer program, characterized in that the computer program is executed by the processor to run the method of any one of claims 1-3.

Images