Seismic data regularization method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention provides a seismic data regularization method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring actual seismic data acquired by an exploration data observation system, the data comprising actual shot point coordinates and travel time data in the exploration data observation system, acquiring coordinates of a target shot point input by a user, constructing a travel time neural network and a speed neural network according to the actual shot point coordinates and the travel time data, and calculating the speed of the target shot point according to the travel time neural network and the speed neural network. A travel time neural network and a speed neural network are established, a travel function equation guidance-based neural network equation is constructed, the travel function equation is used for connecting the travel time neural network and the speed neural network, and according to the travel function equation guidance-based neural network, the actual shot point coordinates and travel time data are input, and neural network training is performed to obtain a trained travel time network and a trained speed network; and according to the trained travel time network, inputting coordinates of a target shot point to obtain travel time data of the target shot point. According to the method, the regularization of the seismic travel time data is realized, and the accuracy of data processing is ensured.

Description

Technical Field

[0001] The present invention relates to the field of seismic exploration technology, and in particular to a seismic data regularization method, device, electronic equipment and storage medium. Background Art

[0002] In the field of seismic data, especially in the process of industrial seismic exploration data processing, the integrity and regularity of data acquisition are crucial. Complete and regular data acquisition can ensure relatively uniform data coverage, which has an improving effect on subsequent stacking and migration imaging.

[0003] However, in actual data collection, especially land seismic data collection, due to various defects in the observation system, the collected data have problems such as missing data, irregular shot points and detection points, etc. At present, in order to address these data defects, the collected data is processed in the data domain, which is data processing based on data features and has poor results.

[0004] In summary, the current seismic data processing method has the technical problem of poor data processing effect. Summary of the invention

[0005] In order to alleviate the technical problem of poor data processing effect in the current seismic data processing method, the embodiments of the present invention provide a seismic data regularization method, device, electronic device and storage medium.

[0006] In a first aspect, an embodiment of the present invention provides a seismic data regularization method, comprising:

[0007] Acquire actual seismic data collected by the exploration data observation system, wherein the actual seismic data includes coordinates and travel time data of actual shot check points in the exploration data observation system;

[0008] Get the coordinates of the target gun inspection point input by the user;

[0009] According to the actual shot check point coordinates and travel time data, a travel time neural network and a velocity neural network are constructed, and a neural network equation guided by an eikonal equation is constructed, wherein the eikonal equation is used to link the travel time neural network and the velocity neural network;

[0010] According to the neural network guided by the eikonal equation, the actual shot check point coordinates and travel time data are input, and the neural network training is performed to obtain the trained travel time network and velocity network;

[0011] According to the trained travel time network, the coordinates of the target shot check point are input to obtain the travel time data of the target shot check point to complete the regularization of the seismic travel time data.

[0012] In some embodiments, the neural network guided by the eikonal equation inputs the actual shot checkpoint coordinates and travel time data, and performs neural network training to obtain a trained travel time network and velocity network, including:

[0013] Constructing a loss function according to the travel time data error of the travel time network and the eikonal equation of the objective equation;

[0014] The neural network is trained according to the loss function and the actual seismic data to obtain a trained travel time network and a velocity network.

[0015] In some embodiments, the loss function comprises an augmented Lagrangian function.

[0016] In some embodiments, the velocity network and the travel time network are physical information neural networks with the same network architecture.

[0017] In some embodiments, the velocity network and the travel time network are physical information neural networks with different network architectures.

[0018] In some embodiments, the velocity network comprises 7 hidden layers, with 20 neurons in each layer.

[0019] In some embodiments, the travel time network comprises 9 hidden layers, with 25 neurons in each layer.

[0020] In a second aspect, an embodiment of the present invention provides a seismic data regularization device, comprising:

[0021] A first acquisition module is used to acquire actual seismic data collected by the exploration data observation system, wherein the seismic data includes coordinates and travel time data of actual shot check points in the exploration data observation system;

[0022] The second acquisition module is used to obtain the coordinates of the target gun inspection point input by the user;

[0023] A network construction module, used to construct a travel time neural network and a velocity neural network according to the actual shot check point coordinates and travel time data, and to construct a neural network equation guided by an eikonal equation, wherein the eikonal equation is used to link the travel time neural network and the velocity neural network;

[0024] A network training module is used to input the coordinates of the actual shot checkpoints and travel time data according to the neural network guided by the eikonal equation, and perform neural network training to obtain a trained travel time network and velocity network;

[0025] The regularization module is used to input the coordinates of the target shot check point according to the trained travel time network, obtain the travel time data of the target shot check point, so as to complete the regularization of the seismic travel time data.

[0026] In a third aspect, an embodiment of the present invention provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the method described in the first aspect is implemented.

[0027] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by one or more processors, the method described in the first aspect is implemented.

[0028] Compared with the prior art, one or more embodiments of the present invention can at least bring the following beneficial effects:

[0029] The embodiment of the present invention provides a method, device, electronic device and storage medium for regularizing seismic data; the method comprises: obtaining actual seismic data collected by an exploration data observation system, wherein the actual seismic data comprises coordinates and travel time data of actual shot checkpoints in the exploration data observation system; obtaining coordinates of a target shot checkpoint input by a user; constructing a travel time neural network and a velocity neural network according to the coordinates and travel time data of the actual shot checkpoints, and constructing a neural network equation guided by an eikonal equation, wherein the eikonal equation is used to link the travel time neural network with the velocity neural network; inputting the coordinates and travel time data of the actual shot checkpoints into the neural network guided by the eikonal equation, and performing neural network training to obtain a trained travel time network and a velocity network; inputting the coordinates of the target shot checkpoint according to the trained travel time network, and obtaining the travel time data of the target shot checkpoint, so as to complete regularization of seismic travel time data. In the method provided in this scheme, a neural network based on physical guidance is constructed and trained to obtain a trained travel time network and velocity network. Then, based on the trained neural network, the coordinates of the target gun checkpoint are input to obtain the travel time data corresponding to the target gun checkpoint, thereby realizing the regularization of the seismic travel time data. This method ensures the accuracy of data processing and improves the data processing effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0031] Figure 1 A first flow chart of a seismic data regularization method provided in an embodiment of the present application;

[0032] Figure 2A schematic diagram of the distribution of shot points provided in the embodiment of the present application;

[0033] Figure 3 A schematic diagram of the actual travel time data of the test shot gather provided in the embodiment of the present application;

[0034] Figure 4 A schematic diagram of reconstructed travel time data obtained after neural network inversion provided in an embodiment of the present application;

[0035] Figure 5 A schematic diagram of the error distribution between the regularized data and the original data provided in the embodiment of the present application;

[0036] Figure 6 A schematic diagram of the effect of converting the original data, the regularized data, and the error between the original data and the interpolated data into a one-dimensional array and then drawing and displaying it in an embodiment of the present application;

[0037] Figure 7 A schematic diagram of the structure of a seismic data regularization device provided in an embodiment of the present application.

[0038] In the drawings, the same reference numerals are used for the same components, and the drawings are not drawn to scale. DETAILED DESCRIPTION

[0039] The following will describe the implementation methods of the present application in detail with reference to the accompanying drawings and embodiments, so that the implementation process of how the present application applies technical means to solve technical problems and achieve corresponding technical effects can be fully understood and implemented accordingly. The embodiments of the present application and the various features in the embodiments can be combined with each other without conflict, and the technical solutions formed are all within the scope of protection of the present application.

[0040] Meanwhile, in the following description, many specific details are set forth for the purpose of explanation to provide a thorough understanding of the embodiments of the present invention. However, it is obvious to those skilled in the art that the present invention can be implemented without the specific details or the specific manner described herein.

[0041] Embodiment 1

[0042] Figure 1 For a first flow chart of the seismic data regularization method provided in the embodiment of the present application, please refer to Figure 1 , the seismic data regularization method provided in this embodiment includes:

[0043] Step S110: Acquire actual seismic data collected by the exploration data observation system.

[0044] In the present application, the actual seismic data includes the coordinates and travel time data of the actual shot check points in the exploration data observation system.

[0045] Step S120: Obtain the coordinates of the target shot check point input by the user.

[0046] In this embodiment, the coordinates of the target shot check points input by the user according to the data regularization requirements are obtained.

[0047] Step S130: constructing a travel time neural network and a velocity neural network according to the actual shot check point coordinates and travel time data, and constructing a neural network equation guided by an eikonal equation, wherein the eikonal equation is used to link the travel time neural network and the velocity neural network.

[0048] Specifically, a neural network is constructed (as shown in the following formula), which represents the travel time field network (corresponding to the travel time network in the above text) and the underground velocity field network (corresponding to the velocity network in the above text):

[0049]

[0050] Among them, T is the travel time field, v is the underground velocity field, is the spatial coordinate of the target point / shot check point, N t is the neural network that outputs the travel time field, N v It is a neural network that outputs the underground velocity field. These neural networks can be understood as a composite function.

[0051] This embodiment adopts an independent dual network structure: a velocity field network and a travel time field network. The velocity field contains 7 hidden layers, each with 20 neurons; the travel time field contains 9 hidden layers, each with 25 neurons; the activation function of the neuron uses the leaky-LU function, but at the output end, in order to ensure that the neuron output result is a positive value (physically, travel time and speed cannot be negative values), a sigmoid function is used.

[0052] The travel time field and the velocity field are related by the following eikonal equation:

[0053]

[0054] in, is a partial differential operator, which represents the partial differential in three directions (x, y, z), |.| represents the absolute value, and then:

[0055]

[0056]

[0057] are the coordinates of the shot point, Indicates that at the shot point, the travel time data is 0. v is the velocity field mentioned above, v -2 It is the -2 power of the velocity field.

[0058] Step S140: According to the neural network guided by the eikonal equation, the actual shot check point coordinates and travel time data are input, and the neural network training is performed to obtain the trained travel time network and velocity network.

[0059] In the present application, the neural network includes a travel time network and a speed network, wherein the speed network is used to ensure the training accuracy of the travel time network; the trained neural network, i.e. the trained neural network, refers to updating the weights on the already constructed neural network structure so that the neural network reaches the optimal solution.

[0060] In some embodiments, the neural network guided by the eikonal equation inputs the actual shot check point coordinates and travel time data, and performs neural network training to obtain a trained travel time network and velocity network, including: constructing a loss function based on the travel time data error of the travel time network and the eikonal equation of the target equation; training the neural network based on the loss function and the actual seismic data to obtain a trained travel time network and velocity network.

[0061] In some embodiments, the loss function comprises an augmented Lagrangian function.

[0062] Specifically, in order to fit the travel time data, the following augmented Lagrangian function is constructed:

[0063]

[0064] Among them, i represents the summation index, is a partial differential operator, N r is the number of receivers, is the value of the travel time field at the receiver location, is the observed travel time, N is the number of points in the underground velocity field, This is the correction value given because the travel time is zero at the epicenter.

[0065] The augmented Lagrangian function can be used as a loss function and trained using a standardized neural network. After the training, both the velocity field network and the travel time field network can be obtained at the same time.

[0066] In some embodiments, the velocity network and the travel time network are physical information neural networks with the same network architecture.

[0067] In some embodiments, the velocity network and the travel time network are physical information neural networks with different network architectures.

[0068] Specifically, in this embodiment, the speed network and the travel time network can be the same physical information neural network, or can be two network physical information neural networks constructed independently.

[0069] In some embodiments, the velocity network comprises 7 hidden layers, with 20 neurons in each layer.

[0070] In some embodiments, the travel time network comprises 9 hidden layers, with 25 neurons in each layer.

[0071] Specifically, the activation function of the neuron uses the leaky-LU function, but at the output end, in order to ensure that the output result of the neuron is a positive value (physically speaking, the travel time and speed cannot be negative), the sigmoid function is used.

[0072] Step S150: According to the trained travel time network, the coordinates of the target shot check point are input to obtain the travel time data of the target shot check point to complete the regularization of the seismic travel time data.

[0073] In some embodiments, according to the travel time network in the trained neural network, the coordinate information of the target shot check point input by the user is input, that is, the coordinate information of the target shot check point is input into the trained travel time field network to obtain the travel time data corresponding to the target shot check point.

[0074] Specifically, the trained neural network includes a trained travel time field network; after obtaining the trained neural network, the coordinates of the target shot check point input by the user are obtained, and the coordinates of the target shot check point are input into the trained travel time field network to obtain the travel time data corresponding to the target shot check point to complete the regularization of the seismic travel time data.

[0075] In summary, in the seismic data regularization method provided in the embodiment of the present application, a physically guided neural network is constructed and trained to obtain a trained travel time network and a velocity network. Then, based on the trained neural network, the coordinates of the target shot checkpoint are input to obtain the travel time data corresponding to the target shot checkpoint, thereby realizing the regularization of the seismic travel time data. This method ensures the accuracy of data processing and improves the data processing effect.

[0076] Now, let's take earthquake data as travel time data and data management as data regularization processing as an example for explanation.

[0077] The travel time data regularization algorithm of the present invention is divided into two major steps:

[0078] First, the observed travel time data is screened to obtain the shot check points and inversion data that need regularization processing; then, a physics-guided neural network is established, including a travel time field network and an underground velocity field network, and the inversion data is used to perform inversion training based on underground velocity. After the network training is completed, a trained neural network is obtained. Among them, the neural network is determined by the structure and weights. A trained neural network means that the weights are updated on the already built neural network structure so that the neural network reaches the optimal solution.

[0079] Second, for the target point set by the user or the gun inspection point that needs regularization processing, the coordinates are input into the neural network to obtain the travel time data of the point after regularization processing.

[0080] Specifically, they include:

[0081] Construct neural networks to represent the travel time field network and the underground velocity field network respectively:

[0082]

[0083] Among them, T is the travel time field, v is the underground velocity field, is the spatial coordinate of the target point / shot check point, N t is the neural network that outputs the travel time field, N v It is a neural network that outputs the underground velocity field. These neural networks can be understood as a composite function.

[0084] This embodiment adopts an independent dual network structure: a velocity field network and a travel time field network. The velocity field contains 7 hidden layers, each with 20 neurons; the travel time field contains 9 hidden layers, each with 25 neurons; the activation function of the neuron uses the leaky-LU function, but at the output end, in order to ensure that the neuron output result is a positive value (physically, travel time and speed cannot be negative values), a sigmoid function is used.

[0085] The travel time field and the velocity field are related by the following eikonal equation:

[0086]

[0087] in, is a partial differential operator, which represents the partial differential in three directions (x, y, z), |.| represents the absolute value, and then:

[0088]

[0089]

[0090] are the coordinates of the shot point, Indicates that at the shot point, the travel time data is 0. v is the velocity field mentioned above, v -2 It is the -2 power of the velocity field.

[0091] In order to fit the travel time data, the following augmented Lagrangian function is constructed:

[0092]

[0093] Among them, i represents the summation index, is a partial differential operator, N r is the number of receivers, is the value of the travel time field at the receiver location, is the observed travel time, N is the number of points in the underground velocity field, This is the correction value given because the travel time is zero at the epicenter.

[0094] The augmented Lagrangian function can be used as a loss function and trained with a standardized neural network. After the training, both the velocity field network and the travel time field network can be obtained. Among them, the travel time field network is useful for data regularization. For the target point or gun inspection point specified by the user that needs to be regularized, it can be directly input into the travel time field network to obtain the travel time data of the target point.

[0095] In summary, the present invention comprises the following steps:

[0096] Step 1: Construct a neural network to fit the velocity field and travel time field respectively. The velocity field network and the travel time field network can be the same, or two networks can be constructed independently.

[0097] Step 2: Construct an augmented Lagrangian function to fuse the travel time data error and the error of the eikonal equation that controls the travel time, and train the network to obtain the travel time field network;

[0098] Step 3: Input the coordinates of the points specified by the user that need to be regularized into the travel time field network to obtain the travel time data of the target point.

[0099] Based on the above records, this application is now described in detail in combination with specific scenarios.

[0100] In order to verify the seismic data regularization method of the embodiment of the present invention, this case uses a batch of actual data for verification. A shot set is randomly selected as a test shot set; then the remaining shot sets are used for neural network inversion. After the neural network training is completed, the original shot set coordinates are input into the neural network to obtain a reconstructed travel time data, which can be compared with the original data.

[0101] like Figure 2As shown in the figure, the overall distribution of the gun set is relatively scattered. The circle marks a randomly selected gun set, which is used as a test gun set and is not used in the inversion.

[0102] Figure 3 It is a schematic diagram of the real travel time data of the test shot set provided in the embodiment of the present invention; in the training of the neural network, this test shot set does not participate in the training, but is used for comparison with the regularized data obtained after the neural network training.

[0103] Figure 4 It is a schematic diagram of the reconstructed travel time data obtained after the neural network inversion provided in the embodiment of the present application; in this embodiment, after completing the neural network inversion, the coordinates of the test shot set (i.e., the original data coordinates) are input into the neural network to obtain the reconstructed travel time data, wherein the purpose of inputting the coordinates of the test shot set is to compare with the original travel time data.

[0104] Figure 5 is a schematic diagram of the error distribution between the regularized data and the original data provided in the embodiment of the present application; Figure 5 It can be seen that the error between the regularized data and the original data at all positions is within 0.08 seconds.

[0105] In order to compare the data errors more intuitively, Figure 6 Yes Figure 2 , Figure 3 , Figure 4 The original data, the regularized data, and the error between the original data and the interpolated data are converted into one-dimensional arrays and then plotted; Figure 6 As shown in the figure, the first line is the original data, the horizontal axis is the number of receivers, and the vertical axis is the travel time; the second line is the regularized data; the third line is the error between the original data and the interpolated data; Figure 6 It can be seen that the maximum error is around 0.08 seconds, and most errors are within 0.03 seconds.

[0106] In summary, the method provided in this embodiment constructs a neural network based on physical guidance, performs training to obtain a trained travel time network and a velocity network, and then inputs the coordinates of the target gun check point based on the trained neural network to obtain the travel time data corresponding to the target gun check point, thereby realizing the regularization of the seismic travel time data. This method ensures the accuracy of data processing and improves the data processing effect.

[0107] Embodiment 2

[0108] Figure 7 For a structural diagram of the seismic data regularization device provided in the embodiment of the present application, please refer to Figure 7, the seismic data regularization device provided in this embodiment includes:

[0109] A first acquisition module 710 is used to acquire actual seismic data collected by the exploration data observation system, wherein the seismic data includes coordinates and travel time data of actual shot check points in the exploration data observation system;

[0110] The second acquisition module 720 is used to acquire the coordinates of the target shot inspection point input by the user;

[0111] A network construction module 730 is used to construct a travel time neural network and a velocity neural network according to the actual shot check point coordinates and travel time data, and to construct a neural network equation guided by an eikonal equation, wherein the eikonal equation is used to link the travel time neural network and the velocity neural network;

[0112] A network training module 740 is used to perform neural network training according to the neural network guided by the eikonal equation, input the actual shot check point coordinates and travel time data, and obtain a trained travel time network and velocity network;

[0113] The regularization module 750 is used to input the coordinates of the target shot check point according to the trained travel time network, obtain the travel time data of the target shot check point, and complete the regularization of the seismic travel time data.

[0114] In some embodiments, the network training module 740 is also used to construct a loss function based on the travel time data error of the travel time network and the eikonal equation of the target equation; and train the neural network based on the loss function and the actual seismic data to obtain a trained travel time network and velocity network.

[0115] The specific implementation examples based on the above-mentioned modules for executing the method have been described in detail in the first embodiment and will not be repeated here.

[0116] In summary, the seismic data regularization device provided in the embodiment of the present invention constructs a neural network based on physical guidance, performs training to obtain a trained travel time network and a velocity network, and then inputs the coordinates of the target shot check point based on the trained neural network to obtain the travel time data corresponding to the target shot check point, thereby realizing the regularization of the seismic travel time data. This method ensures the accuracy of data processing and improves the data processing effect.

[0117] Embodiment 3

[0118] This embodiment provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the seismic data regularization method described in Embodiment 1 is implemented. It is understood that the electronic device may also include an input / output (I / O) interface and a communication component.

[0119] The processor is used to execute all or part of the steps in the seismic data regularization method in Example 1. The memory is used to store various types of data, which may include instructions of any application or method in the terminal device, and data related to the application.

[0120] The processor can be an application specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field programmable gate array (FPGA), a controller, a microcontroller, a microprocessor or other electronic components, and is used to execute the seismic data regularization method in the above-mentioned embodiment 1.

[0121] The memory can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0122] Embodiment 4

[0123] This embodiment also provides a computer-readable storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory (for example, an SD or DX memory, etc.), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a disk, an optical disk, a server, an App application store, etc., wherein a computer program is stored on the storage medium, and when the computer program is executed by a processor, the following method steps can be implemented:

[0124] Acquire actual seismic data collected by the exploration data observation system, wherein the actual seismic data includes coordinates and travel time data of actual shot check points in the exploration data observation system;

[0125] Get the coordinates of the target gun inspection point input by the user;

[0126] According to the actual shot check point coordinates and travel time data, a travel time neural network and a velocity neural network are constructed, and a neural network equation guided by an eikonal equation is constructed, wherein the eikonal equation is used to link the travel time neural network and the velocity neural network;

[0127] According to the neural network guided by the eikonal equation, the actual shot check point coordinates and travel time data are input, and the neural network training is performed to obtain the trained travel time network and velocity network;

[0128] According to the trained travel time network, the coordinates of the target shot check point are input to obtain the travel time data of the target shot check point to complete the regularization of the seismic travel time data.

[0129] The specific implementation process of the above method steps can be found in Example 1, and this embodiment will not be repeated here.

[0130] Embodiment 5

[0131] The present application also provides a computer program product or a computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the seismic data regularization method described in any of the above embodiments.

[0132] In summary, the embodiments of the present invention provide a method, device, electronic device and storage medium for regularizing seismic data; the method comprises: obtaining actual seismic data collected by an exploration data observation system, wherein the actual seismic data comprises actual shot checkpoint coordinates and travel time data in the exploration data observation system; obtaining the coordinates of a target shot checkpoint input by a user; constructing a travel time neural network and a velocity neural network according to the actual shot checkpoint coordinates and travel time data, and constructing a neural network equation guided by an eikonal equation, wherein the eikonal equation is used to link the travel time neural network and the velocity neural network; according to the neural network guided by the eikonal equation, inputting the actual shot checkpoint coordinates and travel time data, and performing neural network training to obtain a trained travel time network and velocity network; according to the trained travel time network, inputting the coordinates of the target shot checkpoint, and obtaining the travel time data of the target shot checkpoint, so as to complete the regularization of seismic travel time data. In the method provided in this scheme, a neural network based on physical guidance is constructed and trained to obtain a trained travel time network and velocity network. Then, based on the trained neural network, the coordinates of the target gun checkpoint are input to obtain the travel time data corresponding to the target gun checkpoint, thereby realizing the regularization of the seismic travel time data. This method ensures the accuracy of data processing and improves the data processing effect.

[0133] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the pros and cons of the implementation.

[0134] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0135] Although the implementation methods disclosed in this application are as above, the contents described are only implementation methods adopted for facilitating the understanding of this application, and are not intended to limit this application. Any technician in the technical field to which this application belongs can make any modifications and changes in the form and details of implementation without departing from the spirit and scope disclosed in this application, but the scope of patent protection of this application shall still be based on the scope defined in the attached claims.

Claims

1. A seismic data regularization method, characterized in that: include: Acquire actual seismic data collected by the exploration data observation system, wherein the actual seismic data includes coordinates and travel time data of actual shot check points in the exploration data observation system; Get the coordinates of the target gun inspection point input by the user; According to the actual shot check point coordinates and travel time data, a travel time neural network and a velocity neural network are constructed, and a neural network equation guided by an eikonal equation is constructed, wherein the eikonal equation is used to link the travel time neural network and the velocity neural network; According to the neural network guided by the eikonal equation, the actual shot check point coordinates and travel time data are input, and the neural network training is performed to obtain the trained travel time network and velocity network; According to the trained travel time network, the coordinates of the target shot check point are input to obtain the travel time data of the target shot check point to complete the regularization of the seismic travel time data.

2. The seismic data regularization method according to claim 1, characterized in that: The neural network guided by the eikonal equation inputs the actual shot check point coordinates and travel time data, and performs neural network training to obtain a trained travel time network and velocity network, including: Constructing a loss function according to the travel time data error of the travel time network and the eikonal equation of the objective equation; The neural network is trained according to the loss function and the actual seismic data to obtain a trained travel time network and a velocity network.

3. The seismic data regularization method according to claim 2, characterized in that: The loss function includes an augmented Lagrangian function.

4. The seismic data regularization method according to any one of claims 1 to 3, characterized in that: The speed network and the travel time network are physical information neural networks with the same network architecture.

5. The seismic data regularization method according to any one of claims 1 to 3, characterized in that: The speed network and the travel time network are physical information neural networks with different network architectures.

6. The seismic data regularization method according to claim 5, characterized in that: The velocity network contains 7 hidden layers, each with 20 neurons.

7. The seismic data regularization method according to claim 5, characterized in that: The travel time network comprises 9 hidden layers, each with 25 neurons.

8. A seismic data regularization device, characterized in that: include: A first acquisition module is used to acquire actual seismic data collected by the exploration data observation system, wherein the seismic data includes coordinates and travel time data of actual shot check points in the exploration data observation system; The second acquisition module is used to obtain the coordinates of the target gun inspection point input by the user; A network construction module, used to construct a travel time neural network and a velocity neural network according to the actual shot check point coordinates and travel time data, and to construct a neural network equation guided by an eikonal equation, wherein the eikonal equation is used to link the travel time neural network and the velocity neural network; A network training module is used to input the coordinates of the actual shot checkpoints and travel time data according to the neural network guided by the eikonal equation, and perform neural network training to obtain a trained travel time network and velocity network; The regularization module is used to input the coordinates of the target shot check point according to the trained travel time network, obtain the travel time data of the target shot check point, so as to complete the regularization of the seismic travel time data.

9. An electronic device, characterized in that: The method comprises a memory and a processor; a computer program is stored in the memory, and when the computer program is executed by the processor, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by one or more processors, the method according to any one of claims 1 to 7 is implemented.

Images