WO2022062508A1

WO2022062508A1 - Static correction processing method and apparatus for seismic data, and computer device and storage medium

Info

Publication number: WO2022062508A1
Application number: PCT/CN2021/102018
Authority: WO
Inventors: 亢永敢; 魏嘉; 陈金焕; 朱海伟; 庞锐
Original assignee: 中国石油化工股份有限公司; 中国石油化工股份有限公司石油物探技术研究院
Priority date: 2020-09-28
Filing date: 2021-06-24
Publication date: 2022-03-31
Also published as: CN114428334A

Abstract

Provided are a static correction processing method and apparatus for seismic data, and a computer device and a storage medium. The method comprises: acquiring target seismic data; inputting the target seismic data into a pre-trained first-break picking neural network model for training, so as to obtain first-break picking data that contains first first-break time data; inputting the first-break picking data that contains the first first-break time data into a static correction processing neural network for training, so as to obtain first-break picking data and second first-break time data that have been subjected to static correction processing; obtaining, by means of calculation, a static correction amount according to the second first-break time data and the first first-break time data that have been subjected to static correction processing; and correcting seismic data according to the static correction amount. Automatic first-break picking and direct static correction calculation are thus realized. In the method, manual first-break data picking is not required, such that a complex near-surface modeling process is avoided, and an efficient and accurate static correction processing function is realized.

Description

地震数据静校正处理方法、装置、计算机设备和存储介质Seismic data static correction processing method, device, computer equipment and storage medium

相关申请的交叉引用CROSS-REFERENCE TO RELATED APPLICATIONS

本申请要求享有2020年09月28日提交的名称为“地震数据静校正处理方法、装置、计算机设备和存储介质”的中国专利申请CN202011042565.8的优先权，其全部内容通过引用并入本文中。This application claims the priority of Chinese patent application CN202011042565.8, which was filed on September 28, 2020 and entitled "Method, Apparatus, Computer Equipment and Storage Medium for Static Correction of Seismic Data", the entire contents of which are incorporated herein by reference. .

技术领域technical field

本申请涉及地震数据处理技术领域，特别是涉及地震数据静校正处理方法、装置、计算机设备和存储介质。The present application relates to the technical field of seismic data processing, and in particular, to a method, apparatus, computer equipment and storage medium for static correction processing of seismic data.

背景技术Background technique

静校正是地震数据处理的重要环节，静校正处理是否准确，直接关系到后续的一系列处理的效果。当前广泛应用的静校正处理方法是利用初至数据进行近地表速度层析建模处理，获取近地表速度模型数据，利用速度模型计算起伏地表引起的旅行时差进行静校正。需要准确的初至数据和近地表速度模型数据。初至拾取过程费时费力，近地表速度建模过程复杂，获取精确的近地表速度模型是一个困难的过程。面对山地等复杂地表探区的地震数据，要获取精确的初至数据和近地表速度模型，十分困难，从而影响了静校正的效果。针对复杂地表地震数据的静校正处理方法很多，主要集中在初至拾取和高精度近地表速度建模上，如自动初至拾取方法，层析近地表速度建模方法等被广泛研究和应用，取得了一定的处理效果，但是仍然不能完全解决山地等复杂地表的静校正问题。如何获取精确的静校正处理效果，是山地复杂地区的地震勘探面临的一个难题。Static correction is an important part of seismic data processing. The accuracy of static correction processing is directly related to the effect of a series of subsequent processing. The currently widely used static correction processing method is to use the first arrival data to perform near-surface velocity tomography processing, obtain the near-surface velocity model data, and use the velocity model to calculate the travel time difference caused by the undulating surface for static correction. Accurate first-arrival data and near-surface velocity model data are required. The first-arrival picking process is time-consuming and labor-intensive, the near-surface velocity modeling process is complex, and it is a difficult process to obtain an accurate near-surface velocity model. Facing the seismic data of complex surface exploration areas such as mountains, it is very difficult to obtain accurate first-arrival data and near-surface velocity models, which affects the effect of static correction. There are many static correction processing methods for complex surface seismic data, mainly focusing on first-arrival picking and high-precision near-surface velocity modeling, such as automatic first-arrival picking method, tomographic near-surface velocity modeling method, etc. A certain processing effect has been achieved, but the static correction problem of complex surfaces such as mountains still cannot be completely solved. How to obtain accurate static correction processing effect is a difficult problem for seismic exploration in complex mountainous areas.

发明内容SUMMARY OF THE INVENTION

本申请提供地震数据静校正处理方法、装置、计算机设备和存储介质，实现了初至自动拾取和直接静校正计算，且无需人工拾取初至数据，避免了复杂近地表建模过程，实现了高效准确的静校正处理功能。The present application provides a static correction processing method, device, computer equipment and storage medium for seismic data, which realizes the automatic first-arrival picking and direct static correction calculation, and does not need to manually pick the first-arrival data, avoids the complex near-surface modeling process, and realizes high-efficiency Accurate static correction processing function.

第一方面，本申请提供一种地震数据静校正处理方法，包括：In a first aspect, the present application provides a static correction processing method for seismic data, including:

获取目标地震数据；Obtain target seismic data;

将所述目标地震数据输入至预先训练的初至拾取神经网络模型进行训练，获得包含第一初至时间数据的初至拾取数据；Inputting the target seismic data into a pre-trained first-arrival picking neural network model for training to obtain first-arriving picking data including first first-arrival time data;

将包含所述第一初至时间数据的所述初至拾取数据输入至静校正处理神经网络进行训练，获得静校正处理后的初至拾取数据以及第二初至时间数据；Inputting the first-arrival pickup data including the first first-arrival time data into a static correction processing neural network for training to obtain first-arrival pickup data and second first-arrival time data after static correction processing;

根据静校正处理后的所述第二初至时间数据和所述第一初至时间数据，计算得到静校正量；Calculate the static correction amount according to the second first arrival time data and the first first arrival time data after the static correction processing;

根据所述静校正量对地震数据进行校正。The seismic data is corrected according to the static correction amount.

第二方面，本申请提供一种地震数据静校正处理装置，包括：In a second aspect, the present application provides a seismic data static correction processing device, comprising:

目标地震数据获取模块，被配置为获取目标地震数据；a target seismic data acquisition module, configured to acquire target seismic data;

第一神经网络训练模块，被配置为将所述目标地震数据输入至预先训练的初至拾取神经网络模型进行训练，获得包含第一初至时间数据的初至拾取数据；a first neural network training module, configured to input the target seismic data into a pre-trained first-arrival picking neural network model for training, and obtain first-arriving picking data including first first-arrival time data;

第二神经网络训练模块，被配置为将包含所述第一初至时间数据的所述初至拾取数据输入至静校正处理神经网络进行训练，获得静校正处理后的初至拾取数据以及第二初至时间数据；The second neural network training module is configured to input the first-arrival pickup data including the first first-arrival time data into a static correction processing neural network for training, and obtain the static-corrected first-arrival pickup data and the second first arrival time data;

静校正量计算模块，被配置为根据静校正处理后的所述第二初至时间数据和所述第一初至时间数据，计算得到静校正量；a static correction amount calculation module, configured to calculate a static correction amount according to the second first arrival time data and the first first arrival time data after the static correction processing;

静校正模块，被配置为根据所述静校正量对地震数据进行校正。A static correction module configured to correct the seismic data according to the static correction amount.

第三方面，本申请提供一种计算机设备，包括存储器和处理器，所述存储器存储有计算机程序，其中，所述处理器执行所述计算机程序时实现以下步骤：In a third aspect, the present application provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, wherein the processor implements the following steps when executing the computer program:

获取目标地震数据；Obtain target seismic data;

第四方面，本申请提供一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序被处理器执行时实现以下步骤：In a fourth aspect, the present application provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

获取目标地震数据；Obtain target seismic data;

地震数据静校正处理方法、装置、计算机设备和存储介质，实现了初至自动拾取和直接静校正计算，且无需人工拾取初至数据，避免了复杂近地表建模过程，实现了高效准确的静校正处理功能。Seismic data static correction processing method, device, computer equipment and storage medium, realize automatic first arrival picking and direct static correction calculation, and do not need to manually pick first arrival data, avoid complex near-surface modeling process, and realize efficient and accurate static correction. Correction processing function.

附图说明Description of drawings

为了更清楚地说明本申请实施例的技术方案，下面将对实施例中所需要使用的附图作简单地介绍，应当理解，以下附图仅示出了本申请的某些实施例，因此不应被看作是对范围的限定，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他相关的附图。In order to illustrate the technical solutions of the embodiments of the present application more clearly, the following drawings will briefly introduce the drawings that need to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore do not It should be regarded as a limitation of the scope, and for those of ordinary skill in the art, other related drawings can also be obtained according to these drawings without any creative effort.

图1为一个实施例中地震数据静校正处理方法的应用场景示意图；1 is a schematic diagram of an application scenario of a static correction processing method for seismic data in one embodiment;

图2为一个实施例中地震数据静校正处理方法的流程示意图；2 is a schematic flowchart of a method for static correction processing of seismic data in one embodiment;

图3为一个实施例中地震数据静校正处理装置的结构框图；3 is a structural block diagram of an apparatus for static correction processing of seismic data in one embodiment;

图4为一个实施例中计算机设备的内部结构图；Fig. 4 is the internal structure diagram of the computer device in one embodiment;

图5为一个实施例中的地震数据静校正处理方法的实施过程示意图；5 is a schematic diagram of an implementation process of a static correction processing method for seismic data in one embodiment;

图6为一个实施例中的初至拾取效果图；6 is a first-arrival pickup effect diagram in one embodiment;

图7A为一个实施例中的静校正处理前的初至拾取示意图；7A is a schematic diagram of first-arrival pickup before static correction processing in one embodiment;

图7B为一个实施例中的静校正处理后的初至拾取示意图。FIG. 7B is a schematic diagram of first-arrival pickup after static correction processing in one embodiment.

具体实施方式detailed description

下面将结合本申请实施例中附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本申请一部分实施例，而不是全部的实施例。通常在此处附图中描述和示出的本申请实施例的组件可以以各种不同的配置来布置和设计。因此，以下对在附图中提供的本申请的实施例的详细描述并非旨在限制要求保护的本申请的范围，而是仅仅表示本申请的选定实施例。基于本申请的实施例，本领域技术人员在没有做出创造性劳动的前提下所获得的所有其他实施例，都属于本申请保护的范围。The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application generally described and illustrated in the drawings herein may be arranged and designed in a variety of different configurations. Thus, the following detailed description of the embodiments of the application provided in the accompanying drawings is not intended to limit the scope of the application as claimed, but is merely representative of selected embodiments of the application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative work fall within the protection scope of the present application.

应注意到：相似的标号和字母在下面的附图中表示类似项，因此，一旦某一项在一个附图中被定义，则在随后的附图中不需要再对其进行定义和解释。同时，在本申请的描述中，术语“第一”、“第二”等仅被配置为区分描述，而不能理解为指示或暗示相对重要性。It should be noted that like numerals and letters refer to like items in the following figures, so once an item is defined in one figure, it does not need to be defined or explained in subsequent figures. Meanwhile, in the description of the present application, the terms 'first', 'second', etc. are only configured to distinguish the description, and cannot be understood as indicating or implying relative importance.

实施例1Example 1

本申请提供的地震数据静校正处理方法，可以应被配置为如图1所示的应用环境中。其中，计算机102通过网络与服务器104通过网络进行通信。其中，终端102可以但不限于是各种个人计算机、服务器、笔记本电脑、智能手机、平板电脑和便携式可穿戴设备，服务器104可以用独立的服务器或者是多个服务器组成的服务器集群来实现。用户通过终端102将目标地震数据发送至服务器104，服务器104获取目标地震数据；将所述目标地震数据输入至预先训练的初至拾取神经网络模型进行训练，获得包含第一初至时间数据的初至拾取数据；将包含所述第一初至时间数据的所述初至拾取数据输入至静校正处理神经网络进行训练，获得静校正处理后的初至拾取数据以及第二初至时间数据；根据静校正处理后的所述第二初至时间数据和所述第一初至时间数据，计算得到静校正量；根据所述静校正量对地震数据进行校正。The seismic data static correction processing method provided by this application can be configured in the application environment as shown in FIG. 1 . The computer 102 communicates with the server 104 through the network through the network. The terminal 102 can be, but is not limited to, various personal computers, servers, notebook computers, smart phones, tablet computers and portable wearable devices, and the server 104 can be implemented by an independent server or a server cluster composed of multiple servers. The user sends the target seismic data to the server 104 through the terminal 102, and the server 104 obtains the target seismic data; the target seismic data is input into the pre-trained first arrival picking neural network model for training, and the first arrival time data including the first arrival time data is obtained. to pick-up data; input the first-arrival pick-up data including the first first-arrival time data to the static correction processing neural network for training, and obtain the first-arrival pick-up data and the second first-arrival time data after static correction processing; according to The second first arrival time data and the first first arrival time data after the static correction process are calculated to obtain a static correction amount; the seismic data is corrected according to the static correction amount.

实施例2Example 2

本实施例中，如图2所示，提供了一种地震数据静校正处理方法，其包括：In this embodiment, as shown in FIG. 2, a static correction processing method for seismic data is provided, which includes:

步骤210，获取目标地震数据。 Step 210, acquiring target seismic data.

步骤220，将所述目标地震数据输入至预先训练的初至拾取神经网络模型进行训练，获得包含第一初至时间数据的初至拾取数据。Step 220: Input the target seismic data into a pre-trained first-arrival picking neural network model for training, and obtain first-arrival picking data including first first-arrival time data.

具体地，从目标地震数据提取样本地震数据的数据体，去除样本地震数据的文件头描述信息和每道的道头信息。验证数据体的获取是按照地震数据体生成同样大小的数据体，数据体中每个采样点值都设置为0，然后在每一道的初至时间对应的采样点位置的值设置为1，表示这一道数据的初至时间位置。Specifically, the data volume of the sample seismic data is extracted from the target seismic data, and the file header description information of the sample seismic data and the trace header information of each track are removed. The acquisition of the verification data volume is to generate a data volume of the same size according to the seismic data volume. The value of each sampling point in the data volume is set to 0, and then the value of the sampling point position corresponding to the first arrival time of each track is set to 1, indicating that The first arrival time position of this data.

将获取到的地震数据体，一次提取一道数据，将一道的每个采样点的振幅值作为输入数据输入到初至拾取神经网络模型的每个节点中，将输入道对应的含有初至时间的道数据作为输出的检验数据。地震数据在训练的时候，不是按顺序输入，而是间隔一定距离输入一道数据，依次循环，最终完成所有输入数据的训练。The acquired seismic data volume is extracted one data at a time, and the amplitude value of each sampling point of one channel is input as input data into each node of the first-arrival picking neural network model, and the corresponding input channel contains the first-arrival time. The track data is used as the output inspection data. During training, the seismic data is not input in sequence, but input a data at a certain distance, and loops in turn, and finally completes the training of all input data.

具体地，将需要进行初至拾取的地震数据提取出地震道数据体，按道顺序输入到初至拾取神经网络模型中。经过初至拾取神经网络模型计算后，输出层的节点值为1的节点对应的地震道的采样位置作为目标道的初至时间。对所有需要初至拾取的地震数据依次输入道神经网络中，最终获取所有数据的初至时间。Specifically, a seismic trace data volume is extracted from the seismic data that needs first-arrival picking, and is input into the first-arrival picking neural network model according to the track sequence. After the first arrival picking neural network model calculation, the sampling position of the seismic trace corresponding to the node whose node value of the output layer is 1 is taken as the first arrival time of the target trace. All the seismic data that needs to be picked up by the first arrival are input into the neural network in turn, and finally the first arrival time of all the data is obtained.

通过向初至拾取神经网络模型输入一道地震数据，通过初至拾取神经网络模型计算后，输出层输出一道与输入地震道数据样点数一致的数据，数据中每个样点值为1或者0。1表示样点为初至时间点，0表示样点为非初至时间点。本实施例中，该采样点对应的初至时间点即为第一初至时间数据，并且，该第一初至时间数据即为实际高程拾取得到的初至时间。By inputting a piece of seismic data to the first-arrival picking neural network model, and after calculating through the first-arrival picking neural network model, the output layer outputs a piece of data that is consistent with the input seismic trace data samples, and the value of each sample point in the data is 1 or 0. 1 indicates that the sample point is the first arrival time point, and 0 indicates that the sample point is not the first arrival time point. In this embodiment, the first arrival time point corresponding to the sampling point is the first first arrival time data, and the first first arrival time data is the first arrival time obtained by picking up the actual elevation.

步骤230，将包含所述第一初至时间数据的所述初至拾取数据输入至静校正处理神经网络进行训练，获得静校正处理后的初至拾取数据以及第二初至时间数。Step 230: Input the first-arrival pickup data including the first first-arrival time data into a static correction processing neural network for training, and obtain the first-arrival pickup data and the second first-arrival time number after static correction processing.

静校正处理神经网络的目标是建立地表高程与初至时间的关系。神经网络结构分为输入层、中间层和输出层。输入层为地震道数据的炮点位置坐标和高程以及检波点位置的坐标和高程。因此输入层设置六个节点，分别对应每道数据的炮点坐标和高程(Sx，Sy，Sz)，检波点坐标和高程(Rx，Ry，Rz)。中间层设置两层，每层节点数为50。输出层一个节点，输出初至时间。神经网络为全连接网络。The goal of the static correction processing neural network is to establish the relationship between the surface elevation and the time of first arrival. The neural network structure is divided into input layer, middle layer and output layer. The input layer is the coordinates and elevation of the shot position and the coordinates and elevation of the receiver position of the seismic trace data. Therefore, six nodes are set in the input layer, corresponding to the shot coordinates and elevations (Sx, Sy, Sz), and the receiver coordinates and elevations (Rx, Ry, Rz) of each data, respectively. There are two layers in the middle layer, and the number of nodes in each layer is 50. The output layer is a node that outputs the first arrival time. The neural network is a fully connected network.

本实施例中，提取地震数据每一道的炮点坐标和高程(Sx，Sy，Sz)，检波点坐标和高程(Rx，Ry，Rz)，将提取的六个参数输入到经网络的输入层，将输入道的初至时间作为输出层的验证数据。依次对目标地震数据的每一道进行神经网络训练。In this embodiment, the shot coordinates and elevations (Sx, Sy, Sz) and the receiver coordinates and elevations (Rx, Ry, Rz) of each track of the seismic data are extracted, and the extracted six parameters are input into the input layer of the network. , the first arrival time of the input channel is used as the verification data of the output layer. Neural network training is performed on each track of the target seismic data in turn.

步骤240，根据静校正处理后的所述第二初至时间数据和所述第一初至时间数据，计算得到静校正量。Step 240: Calculate a static correction amount according to the second first arrival time data and the first first arrival time data after the static correction processing.

在一个实施例中，所述根据静校正处理后的所述第二初至时间数据和所述第一初至时间数据，计算得到静校正量的步骤包括：计算静校正处理后的所述第二初至时间数据和所述第一初至时间数据之差，得到所述静校正量。In one embodiment, the step of calculating the static correction amount according to the second first arrival time data and the first first arrival time data after static correction processing includes: calculating the first arrival time after static correction processing. The difference between the second first arrival time data and the first first arrival time data is to obtain the static correction amount.

训练完成后的神经网络建立了每道高程数据与初至时间之间的关系，对需要静校正处理的目标地震数据重置静校正的高程，输入到神经网络中，经过神经网络计算输出的值为目标高程的初至时间。利用目标高程的初至时间与实际高程拾取的初至时间相减，获得的差值就是目标高程的静校正量。After the training is completed, the neural network establishes the relationship between the elevation data of each track and the first arrival time, resets the static correction elevation for the target seismic data that needs static correction processing, inputs it into the neural network, and calculates the output value through the neural network. is the first arrival time of the target elevation. The first arrival time of the target elevation is subtracted from the first arrival time picked up by the actual elevation, and the difference obtained is the static correction amount of the target elevation.

步骤250，根据所述静校正量对地震数据进行校正。Step 250: Correct the seismic data according to the static correction amount.

对每一道数据按照以上操作，获取目标高程的静校正，完成静校正处理。According to the above operations for each piece of data, the static correction of the target elevation is obtained, and the static correction processing is completed.

上述实施例中，实现了初至自动拾取和直接静校正计算。本方法无需人工拾取初至数据，避免了复杂近地表建模过程，实现了高效准确的静校正处理功能。In the above-mentioned embodiment, first-arrival automatic pick-up and direct static correction calculation are realized. The method does not need to manually pick up the first arrival data, avoids the complex near-surface modeling process, and realizes an efficient and accurate static correction processing function.

在一个实施例中，所述将所述目标地震数据输入至预先训练的初至拾取神经网络模型进行训练，获得包含初至时间的初至拾取数据的步骤之前包括：In one embodiment, before the step of inputting the target seismic data into a pre-trained first-arrival picking neural network model for training, the step of obtaining the first-arriving picking data including the first-arrival time includes:

获取预设格式的样本数据，其中，所述预设格式的样本数据为炮集数据的每一道数据，地震道数据的格式包括文件头、每道的道头数据和道数据体，所述道数据体记录每个采样点上的振幅值；将所述预设格式的样本数据输入至初至拾取神经网络进行训练，得到所述初至拾取神经网络模型。Obtain sample data in a preset format, wherein the sample data in the preset format is each trace of shot collection data, and the format of the seismic trace data includes a file header, trace header data for each trace, and a trace data body. The data volume records the amplitude value at each sampling point; the sample data in the preset format is input into the first-arrival picking neural network for training, and the first-arrival picking neural network model is obtained.

具体地，深度神经网络结构设计完成后，需要用样本数据对神经网络模型进行训练，以获得初至拾取神经网络模型。样本数据采用人工拾取得到的准确初至时间数据。输入样本为炮集数据的每一道数据，地震道数据的格式包括文件头、每道的道头数据和道数据体。道数据体记录每个采样点上的振幅值，神经网络输入数据只需要道数据体，因此需要对输入地震数据进行重构，剥离文件头描述数据和每道的道头数据，保留每道的道数据体，组成纯数据体的样本数据。输出样本数据根据输入的样本数据和初至时间构建。根据输入样本数据的大小，生成一个相同大小的空数据体，数据体中每个采样点的值为0。根据初至时间，将输出样本数据中每道数据的对应采样点的值设置为1，得到输出样本数据。神经网络的训练过程是：将生成的输入样本数据按照道顺序，每次输入一道数据，输入到神经网络的输入层，输入道数据的每个采样点对应一个输入层的节点。输出数据为对应道的输出样本数据。每个输出样本的采样点对应一个输出层的节点。Specifically, after the design of the deep neural network structure is completed, the neural network model needs to be trained with sample data to obtain the first-arrival picking neural network model. The sample data adopts the accurate first arrival time data obtained by manual picking. The input sample is each track data of shot set data, and the format of seismic track data includes file header, track header data of each track and track data body. The trace data body records the amplitude value at each sampling point. The neural network input data only needs the trace data body, so it is necessary to reconstruct the input seismic data, strip the file header description data and the trace header data of each trace, and retain the data of each trace. The data body is the sample data that constitutes the pure data body. The output sample data is constructed from the input sample data and first arrival time. According to the size of the input sample data, an empty data body of the same size is generated, and the value of each sampling point in the data body is 0. According to the first arrival time, the value of the corresponding sampling point of each track of data in the output sample data is set to 1 to obtain the output sample data. The training process of the neural network is as follows: the generated input sample data is input in the order of the channels, and each time the data is input to the input layer of the neural network, each sampling point of the input channel data corresponds to a node of the input layer. The output data is the output sample data of the corresponding channel. The sampling point of each output sample corresponds to a node of the output layer.

在一个实施例中，所述获取预设格式的样本数据的步骤之前还包括：In one embodiment, before the step of acquiring the sample data in the preset format, the step further includes:

构建包括第一输入层、第一中间层和第一输出层的初至拾取神经网络，其中，所述第一输入层被配置为输入一道地震数据，通过所述第一中间层计算后，所述第一输出层输出一道与输入的地震道数据的样点数一致的数据，输出的数据中每个样点值为第一样点值或第二样点值。Constructing a first-arrival picking neural network including a first input layer, a first intermediate layer and a first output layer, wherein the first input layer is configured to input a piece of seismic data, and after calculation by the first intermediate layer, the The first output layer outputs a piece of data that is consistent with the sample number of the input seismic trace data, and each sample point value in the output data is the first sample point value or the second sample point value.

本实施例中，根据地震数据的特点，初至拾取神经网络设置有输入层、中间层和输出层。输入层的节点数与地震数据一道的采样点数一致，被配置为输入一道地震数据。中间层为两层，每层的节点数与输入层一致，输出层的节点数与输入层节点数一致。网络为全连接网络。输入层输入一道地震数据，通过中间层计算后，输出层输出一道与输入地震道数据样点数一致的数据，数据中每个样点值为1或者0。1表示样点为初至时间点，0表示样点为非初至时间点。In this embodiment, according to the characteristics of the seismic data, the first-arrival picking neural network is provided with an input layer, an intermediate layer and an output layer. The number of nodes in the input layer is the same as the number of sampling points of the seismic data one, and is configured to input one seismic data. The middle layer consists of two layers, the number of nodes in each layer is the same as that of the input layer, and the number of nodes in the output layer is the same as the number of nodes in the input layer. The network is a fully connected network. A piece of seismic data is input to the input layer. After calculation by the middle layer, the output layer outputs a piece of data that is consistent with the number of sample points of the input seismic trace data. The value of each sample point in the data is 1 or 0. 1 means that the sample point is the first arrival time point. 0 indicates that the sample point is not the first arrival time point.

具体地，首先，对初至拾取神经网络结构进行设置。按照一个输入层，两个中间层，一个输出层设计神经网络结构，输入层的节点数与需要初至拾取的地震数据采样点数一致，中间两层节点数和输出层节点数与输入层节点数一致。随后，生成训练样本数据。选取部分需要进行初至拾取的地震数据作为训练样本数据，首先对训练样本地震数据进行人工拾取初至，获取精确的初至时间。根据神经网络结构，提取样本地震数据的数据体，去除样本地震数据的文件头描述信息和每道的道头信息。验证数据体的获取是按照地震数据体生成同样大小的数据体，数据体中每个采样点值都设置为0，然后在每一道的初至时间对应的采样点位置的值设置为1，表示这一道数据的初至时间位置。然后，确定训练参数。神经网络的训练参数是确定训练效果的关键因素，考虑到训练的计算量和精度，通过循环次数和误差量两个参数控制神经网络训练。训练误差确定训练的精度，防止出现过拟合。循环次数控制训练的计算量，防止陷入多次循环，无法正常结束。Specifically, first, the first-arrival picking neural network structure is set. The neural network structure is designed according to one input layer, two middle layers, and one output layer. The number of nodes in the input layer is the same as the number of seismic data sampling points that need to be picked up first, and the number of nodes in the middle two layers and the number of nodes in the output layer are the same as the number of nodes in the input layer. Consistent. Subsequently, training sample data is generated. Part of the seismic data that needs to be picked for the first arrival is selected as the training sample data. First, the first arrival of the training sample seismic data is manually picked to obtain the accurate first arrival time. According to the neural network structure, the data body of the sample seismic data is extracted, and the file header description information of the sample seismic data and the track header information of each track are removed. The acquisition of the verification data volume is to generate a data volume of the same size according to the seismic data volume. The value of each sampling point in the data volume is set to 0, and then the value of the sampling point position corresponding to the first arrival time of each track is set to 1, indicating that The first arrival time position of this data. Then, the training parameters are determined. The training parameters of the neural network are the key factors to determine the training effect. Considering the calculation amount and accuracy of the training, the training of the neural network is controlled by two parameters: the number of cycles and the amount of error. The training error determines the accuracy of the training and prevents overfitting. The number of loops controls the calculation amount of the training, preventing it from falling into multiple loops and failing to end normally.

在一个实施例中，所述将所述目标地震数据输入至预先训练的初至拾取神经网络模型进行训练，获得包含第一初至时间数据的初至拾取数据的步骤包括：In one embodiment, the step of inputting the target seismic data into a pre-trained first-arrival picking neural network model for training, and obtaining the first-arriving picking data including the first first-arrival time data includes:

将所述目标地震数据输入至预先训练的初至拾取神经网络模型进行训练，所述初至拾取神经网络模型输出包含样点值的所述初至拾取数据，其中，所述样点值包括第一样点值和第二样点值；通过所述初至拾取神经网络模型提取出所述样点值为所述第一样点值的所述初至拾取数据，获取所述第一样点值对应的所述初至拾取数据的所述第一初至时间数据。The target seismic data is input into a pre-trained first-arrival picking neural network model for training, and the first-arrival picking neural network model outputs the first-arrival picking data including sample point values, wherein the sample point values include A sample point value and a second sample point value; extract the first arrival pick-up data whose sample point value is the first sample point value through the first arrival picking neural network model, and obtain the first sample point value of the first first arrival time data of the first arrival pickup data corresponding to the value.

本实施例中，利用训练后的神经网络模型，将需要进行初至拾取处理的目标地震数据按照输入样本的要求，生成输入数据，输入到训练好的神经网络中，神经网络计算后输出的一道数据，提取输出道中采样值为1的样点，获取值为1样点所在的位置时间作为此道的初至拾取结果。In this embodiment, the trained neural network model is used to generate the input data of the target seismic data that needs to be processed for the first-time pick-up according to the requirements of the input samples, and input them into the trained neural network. Data, extract the sample point with the sampling value of 1 in the output channel, and obtain the position time of the sample point with the value of 1 as the first arrival picking result of this channel.

在一个实施例中，所述将包含所述第一初至时间数据的所述初至拾取数据输入至静校正处理神经网络进行训练，获得静校正处理后的初至拾取数据以及第二初至时间数据的步骤包括：In one embodiment, the first-arrival pickup data including the first first-arrival time data is input into a static correction processing neural network for training, and the static-corrected first-arrival pickup data and the second first arrival are obtained The steps for temporal data include:

将包含所述第一初至时间数据的所述初至拾取数据输入至静校正处理神经网络进行训练；通过所述静校正处理神经网络，对需要静校正处理的目标地震数据重置静校正的高程，输出目标高程的初至时间。The first-arrival pickup data including the first first-arrival time data is input into the static correction processing neural network for training; the static correction processing neural network is used to reset the static correction for the target seismic data that needs static correction processing. Elevation, output the first arrival time of the target elevation.

具体地，对已经拾取初至的地震数据提取每一道的炮点坐标和高程(Sx，Sy，Sz)，以及检波点坐标和高程(Rx，Ry，Rz)。将获得的每一道的六个数据分别输入到神经网络的输入层的六个节点，输出验证数据为输入道的初至时间。将所有拾取初至的地震道按照以上处理过程提取训练样本对神经网络进行训练。Specifically, the shot coordinates and elevations (Sx, Sy, Sz) of each track, and the receiver coordinates and elevations (Rx, Ry, Rz) are extracted from the seismic data for which the first arrivals have been picked up. The obtained six data of each track are input to the six nodes of the input layer of the neural network respectively, and the output verification data is the first arrival time of the input track. All first-arrival seismic traces are picked up according to the above process to extract training samples to train the neural network.

在一个实施例中，所述将包含所述第一初至时间数据的所述初至拾取数据输入至静校正处理神经网络进行训练，获得静校正处理后的初至拾取数据以及第二初至时间数据的步骤之前还包括：In one embodiment, the first-arrival pickup data including the first first-arrival time data is input into a static correction processing neural network for training, and the static-corrected first-arrival pickup data and the second first arrival are obtained The steps for time data also include:

构建包括第二输入层、第二中间层和第二输出层的静校正处理神经网络。A static correction processing neural network is constructed including a second input layer, a second intermediate layer, and a second output layer.

具体地，静校正处理神经网络的目标是建立地表高程与初至时间的关系。神经网络结构分为输入层、中间层和输出层。输入层为地震道数据的炮点位置坐标和高程以及检波点位置的坐标和高程。因此输入层设置六个节点，分别对应每道数据的炮点坐标和高程(Sx，Sy，Sz)，检波点坐标和高程(Rx，Ry，Rz)。中间层设置两层，每层节点数为50。输出层一个节点，输出初至时间。神经网络为全连接网络。Specifically, the goal of the static correction processing neural network is to establish the relationship between the surface elevation and the first arrival time. The neural network structure is divided into input layer, middle layer and output layer. The input layer is the coordinates and elevation of the shot position and the coordinates and elevation of the receiver position of the seismic trace data. Therefore, six nodes are set in the input layer, corresponding to the shot coordinates and elevations (Sx, Sy, Sz), and the receiver coordinates and elevations (Rx, Ry, Rz) of each data, respectively. There are two layers in the middle layer, and the number of nodes in each layer is 50. The output layer is a node that outputs the first arrival time. The neural network is a fully connected network.

上述实施例中，利用深度神经网络，对地震数据自动初至拾取，获取初至数据后，利用深度神经网络对初至数据进行处理，获取初至时间与地表高程的关系。利用初至时间与地表高程的关系，通过设置不同目标地表高程，计算相应高程的初至时间，实现了直接静校正处理。In the above embodiment, the deep neural network is used to automatically pick up the first arrivals of the seismic data, and after the first arrivals data are obtained, the deep neural networks are used to process the first arrivals data to obtain the relationship between the first arrivals time and the surface elevation. Using the relationship between the first arrival time and the surface elevation, by setting different target surface elevations, the first arrival time of the corresponding elevation is calculated, and the direct static correction processing is realized.

应该理解的是，虽然图2的流程图中的各个步骤按照箭头的指示依次显示，但是这些步骤并不是必然按照箭头指示的顺序依次执行。除非本文中有明确的说明，这些步骤的执行并没有严格的顺序限制，这些步骤可以以其它的顺序执行。而且，图2中的至少一部分步骤可以包括多个子步骤或者多个阶段，这些子步骤或者阶段并不必然是在同一时刻执行完成，而是可以在不同的时刻执行，这些子步骤或者阶段的执行顺序也不必然是依次进行，而是可以与其它步骤或者其它步骤的子步骤或者阶段的至少一部分轮流或者交替地执行。It should be understood that although the various steps in the flowchart of FIG. 2 are shown in sequence according to the arrows, these steps are not necessarily executed in the sequence shown by the arrows. Unless explicitly stated herein, the execution of these steps is not strictly limited to the order, and these steps may be performed in other orders. Moreover, at least a part of the steps in FIG. 2 may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed and completed at the same time, but may be executed at different times. The execution of these sub-steps or stages The sequence is also not necessarily sequential, but may be performed alternately or alternately with other steps or sub-steps of other steps or at least a portion of a phase.

实施例3Example 3

本申请利用深度神经网络，对地震数据自动初至拾取，获取初至数据后，利用深度神经网络对初至数据进行处理，获取初至时间与地表高程的关系。利用初至时间与地表高程的关系，通过设置不同目标地表高程，计算相应高程的初至时间，实现了直接静校正处理。直接静校正处理过程是：设置统一的地表高程，利用深度神经网络计算统一高程的初至时间，统一高程的初至时间与实际拾取的初至时间相减，获取目标高程的时间校正量，利用时间校正量直接进行静校正处理。The present application uses a deep neural network to automatically pick up the first arrivals of the seismic data, and after acquiring the first arrivals data, the deep neural network is used to process the first arrivals data to obtain the relationship between the first arrivals time and the surface elevation. Using the relationship between the first arrival time and the surface elevation, by setting different target surface elevations, the first arrival time of the corresponding elevation is calculated, and the direct static correction processing is realized. The direct static correction processing process is: set a unified surface elevation, use the deep neural network to calculate the first arrival time of the unified elevation, subtract the first arrival time of the unified elevation from the actual picked first arrival time, obtain the time correction amount of the target elevation, use The time correction amount is directly processed for static correction.

本实施例中，地震数据静校正处理的过程如下：In this embodiment, the process of static correction processing of seismic data is as follows:

(1)初至拾取神经网络结构(1) First pick up neural network structure

根据地震数据的特点，初至拾取神经网络设置有输入层、中间层和输出层。输入层的节点数与地震数据一道的采样点数一致，被配置为输入一道地震数据。中间层为两层，每层的节点数与输入层一致，输出层的节点数与输入层节点数一致。网络为全连接网络。输入层输入一道地震数据，通过中间层计算后，输出层输出一道与输入地震道数据样点数一致的数据，数据中每个样点值为1或者0。1表示样点为初至时间点，0表示样点为非初至时间点。According to the characteristics of seismic data, the first-arrival picking neural network is provided with an input layer, an intermediate layer and an output layer. The number of nodes in the input layer is the same as the number of sampling points of the seismic data one, and is configured to input one seismic data. The middle layer consists of two layers, the number of nodes in each layer is the same as that of the input layer, and the number of nodes in the output layer is the same as the number of nodes in the input layer. The network is a fully connected network. A piece of seismic data is input to the input layer. After calculation by the middle layer, the output layer outputs a piece of data that is consistent with the number of sample points of the input seismic trace data. The value of each sample point in the data is 1 or 0. 1 means that the sample point is the first arrival time point. 0 indicates that the sample point is not the first arrival time point.

(2)初至拾取样本数据获取(2) First pick up sample data acquisition

深度神经网络结构设计完成后，需要用样本数据对神经网络模型进行训练。样本数据采用人工拾取得到的准确初至时间数据。输入样本为炮集数据的每一道数据，地震道数据的格式包括文件头、每道的道头数据和道数据体。道数据体记录每个采样点上的振幅值，神经网络输入数据只需要道数据体，因此需要对输入地震数据进行重构，剥离文件头描述数据和每道的道头数据，保留每道的道数据体，组成纯数据体的样本数据。输出样本数据根据输入的样本数据和初至时间构建。根据输入样本数据的大小，生成一个相同大小的空数据体，数据体中每个采样点的值为0。根据初至时间，将输出样本数据中每道数据的对应采样点的值设置为1，得到输出样本数据。神经网络的训练过程是：将生成的输入样本数据按照道顺序，每次输入一道数据，输入到神经网络的输出层，输入道数据的每个采样点对应一个输入层的节点。输出数据为对应道的输出样本数据。每个输出样本的采样点对应一个输出层的节点。After the design of the deep neural network structure is completed, the neural network model needs to be trained with sample data. The sample data adopts the accurate first arrival time data obtained by manual picking. The input sample is each track data of shot set data, and the format of seismic track data includes file header, track header data of each track and track data body. The trace data body records the amplitude value at each sampling point. The neural network input data only needs the trace data body, so it is necessary to reconstruct the input seismic data, strip the file header description data and the trace header data of each trace, and retain the data of each trace. The data body is the sample data that constitutes the pure data body. The output sample data is constructed from the input sample data and first arrival time. According to the size of the input sample data, an empty data body of the same size is generated, and the value of each sampling point in the data body is 0. According to the first arrival time, the value of the corresponding sampling point of each data in the output sample data is set to 1 to obtain the output sample data. The training process of the neural network is: input the generated input sample data in the order of the channels, input one data at a time, and input it to the output layer of the neural network, and each sampling point of the input data corresponds to a node of the input layer. The output data is the output sample data of the corresponding channel. The sampling point of each output sample corresponds to a node of the output layer.

(3)初至拾取数据获取(3) First-time pickup data acquisition

利用训练后的神经网络模型，将需要进行初至拾取处理的目标地震数据按照输入样本的要求，生成输入数据，输入到训练好的神经网络中，神经网络计算后输出的一道数据，提取输出道中采样值为1的样点，获取值为1样点所在的位置时间作为此道的初至拾取结果。Using the trained neural network model, the target seismic data that needs to be first picked up is generated according to the requirements of the input sample, and the input data is input into the trained neural network. For the sample point with the sampling value of 1, the position time of the sample point with the value of 1 is obtained as the first arrival picking result of this track.

(4)静校正处理神经网络结构设置(4) Static correction processing neural network structure settings

静校正处理神经网络的目标是建立地表高程与初至时间的关系。神经网络结构分为输入层、中间层和输出层。输入层为地震道数据的炮点位置坐标和高程以及检波点位置的坐标和高程。因此输入层设置六个节点，分布对应每道数据的炮点坐标和高程(Sx，Sy，Sz)，检波点坐标和高程(Rx，Ry，Rz)。中间层设置两层，每层节点数为50。输出层一个节点，输出初至时间。神经网络为全连接网络。The goal of the static correction processing neural network is to establish the relationship between the surface elevation and the time of first arrival. The neural network structure is divided into input layer, middle layer and output layer. The input layer is the coordinates and elevation of the shot position and the coordinates and elevation of the receiver position of the seismic trace data. Therefore, six nodes are set in the input layer, and the distribution corresponds to the shot coordinates and elevations (Sx, Sy, Sz), and the receiver coordinates and elevations (Rx, Ry, Rz) of each data. There are two layers in the middle layer, and the number of nodes in each layer is 50. The output layer is a node that outputs the first arrival time. The neural network is a fully connected network.

(5)训练样本生成(5) Training sample generation

对已经拾取初至的地震数据提取每一道的炮点坐标和高程(Sx，Sy，Sz)，以及检波点坐标和高程(Rx，Ry，Rz)。将获得的每一道的六个数据分别输入到神经网络的输入层的六个节点，输出验证数据为输入道的初至时间。将所有拾取初至的地震道按照以上处理过程提取训练样本对神经网络进行训练。The shot coordinates and elevations (Sx, Sy, Sz) of each trace, and the receiver coordinates and elevations (Rx, Ry, Rz) are extracted from the seismic data for which the first arrivals have been picked up. The obtained six data of each track are input to the six nodes of the input layer of the neural network respectively, and the output verification data is the first arrival time of the input track. All first-arrival seismic traces are picked up according to the above process to extract training samples to train the neural network.

(6)静校正处理(6) Static correction processing

训练完成后的神经网络建立了每道高程数据与初至时间之间的关系，对需要静校正处理的目标地震数据重置静校正的高程，输入到神经网络中，经过神经网络计算输出的值为目标高程的初至时间。利用目标高程的初至时间与实际高程拾取的初至时间相减，获得的差值就是目标高程的静校正量。对每一道数据按照以上操作，获取目标高程的静校正，完成静校正处理。After the training is completed, the neural network establishes the relationship between the elevation data of each track and the first arrival time, resets the static correction elevation for the target seismic data that needs static correction processing, inputs it into the neural network, and calculates the output value through the neural network. is the first arrival time of the target elevation. The first arrival time of the target elevation is subtracted from the first arrival time picked up by the actual elevation, and the difference obtained is the static correction amount of the target elevation. According to the above operations for each piece of data, the static correction of the target elevation is obtained, and the static correction processing is completed.

本申请提供一种基于深度神经网络的地震数据静校正方法，实现地震数据直接静校正处理，避免了初至拾取和近地表速度建模等处理过程，满足复杂地表地震数据静校正处理需求，改进地震数据的处理效果，降低了勘探成本，提高了经济效益。The present application provides a static correction method for seismic data based on a deep neural network, which realizes the direct static correction processing of seismic data, avoids processing processes such as first-arrival picking and near-surface velocity modeling, meets the static correction processing requirements of complex surface seismic data, and improves the The processing effect of seismic data reduces exploration costs and improves economic benefits.

实施例4Example 4

请结合图5，第一步，初至拾取神经网络结构设置。按照一个输入层，两个中间层，一个输出层设计神经网络结构，输入层的节点数与需要初至拾取的地震数据采样点数一致，中间两层节点数和输出层点节点数与输入层节点数一致。Please refer to Figure 5, the first step, the first step to pick up the neural network structure settings. The neural network structure is designed according to one input layer, two middle layers, and one output layer. The number of nodes in the input layer is consistent with the number of seismic data sampling points that need to be picked up first, and the number of nodes in the middle two layers and the number of nodes in the output layer are the same as those in the input layer. The numbers are the same.

第二步，训练样本数据生成。选取部分需要进行初至拾取的地震数据作为训练样本数据，首先对训练样本地震数据进行人工拾取初至，获取精确的初至时间。根据神经网络结构，提取样本地震数据的数据体，去除样本地震数据的文件头描述信息和每道的道头信息。验证数据体的获取是按照地震数据体生成同样大小的数据体，数据体中每个采样点值都设置为0，然后在每一道的初至时间对应的采样点位置的值设置为1，表示这一道数据的初至时间位置。The second step is to generate training sample data. Part of the seismic data that needs to be picked for the first arrival is selected as the training sample data. First, the first arrival of the training sample seismic data is manually picked to obtain the accurate first arrival time. According to the neural network structure, the data body of the sample seismic data is extracted, and the file header description information of the sample seismic data and the track header information of each track are removed. The acquisition of the verification data volume is to generate a data volume of the same size according to the seismic data volume. The value of each sampling point in the data volume is set to 0, and then the value of the sampling point position corresponding to the first arrival time of each track is set to 1, indicating that The first arrival time position of this data.

第三步，确定训练参数。神经网络的训练参数是确定训练效果的关键因素，考虑到训练的计算量和精度，通过循环次数和误差量两个参数控制神经网络训练。训练误差确定训练的精度，防止出现过拟合。循环次数控制训练的计算量，防止陷入多次循环，无法正常结束。The third step is to determine the training parameters. The training parameters of the neural network are the key factors to determine the training effect. Considering the calculation amount and accuracy of the training, the training of the neural network is controlled by two parameters: the number of cycles and the amount of error. The training error determines the accuracy of the training and prevents overfitting. The number of loops controls the amount of calculation for training, preventing it from falling into multiple loops and failing to end normally.

第四步，神经网络训练。根据神经网络结构和参数，将第二步获取地震数据体，一次提取一道数据，将一道的每个采样点的振幅值作为输入数据输入到神经网络输入层的每个节点中，将输入道对应的含有初至时间的道数据作为输出的检验数据。地震数据在训练的时候，不是按顺序输入，而是间隔一定距离输入一道数据，依次循环，最终完成所有输入数据的训练。The fourth step is neural network training. According to the structure and parameters of the neural network, the seismic data volume is acquired in the second step, one data is extracted at a time, and the amplitude value of each sampling point of one channel is input into each node of the input layer of the neural network as input data, and the input channel corresponds to The track data containing the first arrival time is used as the output inspection data. During training, the seismic data is not input in sequence, but input a data at a certain distance, and loops in turn, and finally completes the training of all input data.

第五步，初至拾取。利用第四步训练后的神经网络，将需要进行初至拾取的地震数据按照第二步的要求提取地震道数据体，按道顺序输入到神经网络中。经过神经网络计算后，输出层的节点值为1的节点对应的地震道的采样位置作为目标道的初至时间。对所有需要初至拾取的地震数据依次输入到神经网络中，最终获取所有数据的初至时间。拾取后的初至图像如图6所示。The fifth step, first pick up. Using the neural network trained in the fourth step, the seismic data that needs to be picked up for the first time is extracted according to the requirements of the second step, and the seismic trace data volume is input into the neural network in sequence. After the neural network calculation, the sampling position of the seismic trace corresponding to the node whose node value is 1 in the output layer is taken as the first arrival time of the target trace. All the seismic data that needs to be picked up by the first arrival are input into the neural network in turn, and finally the first arrival time of all the data is obtained. The first-arrival image after pickup is shown in Figure 6.

第六步，静校正处理神经网络结构设置。静校正神经网络结构分为输入层、中间层和输出层。输入层设置六个节点，分别对应每道数据的炮点坐标和高程(Sx，Sy，Sz)，检波点坐标和高程(Rx，Ry，Rz)。中间层设置两层，每层节点数为50。输出层一个节点，输出初至时间。神经网络为全连接网络。The sixth step, static correction deals with the neural network structure settings. The static correction neural network structure is divided into input layer, middle layer and output layer. Six nodes are set in the input layer, corresponding to the shot coordinates and elevations (Sx, Sy, Sz), and the receiver coordinates and elevations (Rx, Ry, Rz) of each data. There are two layers in the middle layer, and the number of nodes in each layer is 50. The output layer is a node that outputs the first arrival time. The neural network is a fully connected network.

第七步，静校正神经网络训练。提取地震数据每一道的炮点坐标和高程(Sx，Sy，Sz)，检波点坐标和高程(Rx，Ry，Rz)，将提取的六个参数输入到神经网络的输入层，将输入道的初至时间作为输出层的验证数据。依次对目标地震数据的每一道进行神经网络训练。The seventh step is to statically correct the neural network training. Extract the shot coordinates and elevations (Sx, Sy, Sz) and receiver coordinates and elevations (Rx, Ry, Rz) of each trace of the seismic data, and input the extracted six parameters into the input layer of the neural network. The first arrival time is used as the validation data for the output layer. Neural network training is performed on each track of the target seismic data in turn.

第八步，静校正处理。保存完成训练的神经网络参数。根据静校正需要，确定地表高程，按照新的地表高程替换需要静校正处理的地震数据道中的炮点高程和检波点高程，获取静校正后的炮点坐标(Sx，Sy，Sz)、检波点坐标(Rx，Ry，Rz)，将获取的新高程的六个参数输入道神经网络中，经过神经网络计算输出对应道的初至时间数据。将目标高程计算得到的初至时间与实际高程拾取得到的初至时间相减，得到的差值就是静校正量。依次计算所有道数据，获得最终的静校正量。按照获取的静校正量对目标地震数据进行校正，实现静校正处理。请参见图7A和图7B，分别为静校正处理前的初至拾取静校正处理后的初至拾取。The eighth step, static correction processing. Save the trained neural network parameters. According to the needs of static correction, determine the surface elevation, replace the shot elevation and receiver elevation in the seismic data trace that needs static correction processing according to the new surface elevation, and obtain the static corrected shot coordinates (Sx, Sy, Sz), receiver point Coordinates (Rx, Ry, Rz), input the acquired six parameters of the new elevation into the neural network, and output the first arrival time data of the corresponding track through the neural network calculation. The first arrival time calculated by the target elevation is subtracted from the first arrival time obtained by picking up the actual elevation, and the difference obtained is the static correction amount. Calculate all the track data in turn to obtain the final static correction amount. Correct the target seismic data according to the acquired static correction amount to realize static correction processing. Please refer to FIG. 7A and FIG. 7B , which are respectively the first arrival picking before the static correction processing and the first arrival picking after the static correction processing.

实施例5Example 5

本实施例中，如图3所示，提供一种地震数据静校正处理装置，包括：In this embodiment, as shown in FIG. 3, a seismic data static correction processing device is provided, including:

目标地震数据获取模块310，被配置为获取目标地震数据；a target seismic data acquisition module 310, configured to acquire target seismic data;

第一神经网络训练模块320，被配置为将所述目标地震数据输入至预先训练的初至拾取神经网络模型进行训练330，获得包含第一初至时间数据的初至拾取数据；The first neural network training module 320 is configured to input the target seismic data into a pre-trained first-arrival picking neural network model for training 330 to obtain first-arrival picking data including first first-arrival time data;

第二神经网络训练模块340，被配置为将包含所述第一初至时间数据的所述初至拾取数据输入至静校正处理神经网络进行训练，获得静校正处理后的初至拾取数据以及第二初至时间数据；The second neural network training module 340 is configured to input the first-arrival pickup data including the first first-arrival time data into a static correction processing neural network for training, and obtain the static corrected first-arrival pickup data and the first-arrival pickup data after static correction processing. Second arrival time data;

静校正量计算模块350，被配置为根据静校正处理后的所述第二初至时间数据和所述第一初至时间数据，计算得到静校正量；The static correction amount calculation module 350 is configured to calculate the static correction amount according to the second first arrival time data and the first first arrival time data after the static correction processing;

静校正模块360，被配置为根据所述静校正量对地震数据进行校正。The static correction module 360 is configured to correct the seismic data according to the static correction amount.

在一个实施例中，所述地震数据静校正处理装置还包括：In one embodiment, the seismic data static correction processing device further includes:

样本数据获取模块，被配置为获取预设格式的样本数据，其中，所述预设格式的样本数据为炮集数据的每一道数据，地震道数据的格式包括文件头、每道的道头数据和道数据体，所述道数据体记录每个采样点上的振幅值；The sample data acquisition module is configured to acquire sample data in a preset format, wherein the sample data in the preset format is each track data of the shot collection data, and the format of the seismic track data includes file header, track header data of each track and a track data body, the track data body records the amplitude value at each sampling point;

初至拾取神经网络模型训练获得模块，被配置为将所述预设格式的样本数据输入至初至拾取神经网络进行训练，得到所述初至拾取神经网络模型。The first-arrival picking neural network model training and obtaining module is configured to input the sample data in the preset format into the first-arriving picking neural network for training to obtain the first-arriving picking neural network model.

初至拾取神经网络构建模块，被配置为构建包括第一输入层、第一中间层和第一输出层的初至拾取神经网络，其中，所述第一输入层被配置为输入一道地震数据，通过所述第一中间层计算后，所述第一输出层输出一道与输入的震道数据的样点数一致的数据，输出的数据中每个样点值为第一样点值或第二样点值。a first-arrival picking neural network building module configured to construct a first-arrival picking neural network including a first input layer, a first intermediate layer, and a first output layer, wherein the first input layer is configured to input a piece of seismic data, After being calculated by the first intermediate layer, the first output layer outputs a piece of data that is consistent with the sample number of the input seismic trace data, and the value of each sample point in the output data is the value of the first sample point or the second sample value. point value.

将所述目标地震数据输入至预先训练的初至拾取神经网络模型进行训练，所述初至拾取神经网络模型输出包含样点值的所述初至拾取数据，其中，所述样点值包括第一样点值和第二样点值；The target seismic data is input into a pre-trained first-arrival picking neural network model for training, and the first-arrival picking neural network model outputs the first-arrival picking data including sample point values, wherein the sample point values include The first sample value and the second sample value;

通过所述初至拾取神经网络模型提取出所述样点值为所述第一样点值的所述初至拾取数据，获取所述第一样点值对应的所述初至拾取数据的所述第一初至时间数据。The first-arrival picking data whose sample point value is the first sample point value is extracted through the first-arrival picking neural network model, and all the first-arriving picking data corresponding to the first sample point value are obtained. Describe the first arrival time data.

在一个实施例中，所述第二神经网络训练模块包括：In one embodiment, the second neural network training module includes:

初至拾取数据输入单元，被配置为将包含所述第一初至时间数据的所述初至拾取数据输入至静校正处理神经网络进行训练；a first-arrival pickup data input unit, configured to input the first-arrival pickup data including the first first-arrival time data to a static correction processing neural network for training;

目标高程输出单元，被配置为通过所述静校正处理神经网络，对需要静校正处理的目标地震数据重置静校正的高程，输出目标高程的初至时间。The target elevation output unit is configured to reset the statically corrected elevation for the target seismic data requiring static correction processing through the static correction processing neural network, and output the first arrival time of the target elevation.

静校正处理神经网络构建模块，被配置为构建包括第二输入层、第二中间层和第二输出层的静校正处理神经网络。A statics processing neural network building block configured to build a statics processing neural network including a second input layer, a second intermediate layer, and a second output layer.

在一个实施例中，所述静校正量计算模块还被配置为计算静校正处理后的所述第二初至时间数据和所述第一初至时间数据之差，得到所述静校正量。In one embodiment, the static correction amount calculation module is further configured to calculate the difference between the second first arrival time data and the first first arrival time data after static correction processing to obtain the static correction amount.

关于地震数据静校正处理装置的具体限定可以参见上文中对于地震数据静校正处理方法的限定，在此不再赘述。上述地震数据静校正处理装置中的各个单元可全部或部分通过软件、硬件及其组合来实现。上述各单元可以硬件形式内嵌于或独立于计算机设备中的处理器中，也可以以软件形式存储于计算机设备中的存储器中，以便于处理器调用执行以上各个单元对应的操作。For the specific limitations of the seismic data static correction processing device, reference may be made to the limitations on the seismic data static correction processing method above, which will not be repeated here. Each unit in the above-mentioned seismic data static correction processing apparatus can be implemented in whole or in part by software, hardware and combinations thereof. The above units may be embedded in or independent of the processor in the computer device in the form of hardware, or may be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above units.

实施例6Example 6

本实施例中，提供了计算机设备。其内部结构图可以如图4所示。该计算机设备包括通过***总线连接的处理器、存储器、网络接口、显示屏和输入装置。其中，该计算机设备的处理器被配置为提供计算和控制能力。该计算机设备的存储器包括非易失性存储介质、内存储器。该非易失性存储介质存储有操作***和计算机程序，且该非易失性存储介质部署有数据库，该数据库被配置为存储了初至拾取神经网络模型和静校正处理神经网络模型。该内存储器为非易失性存储介质中的操作***和计算机程序的运行提供环境。该计算机设备的网络接口被配置为与其他计算机设备通信。该计算机程序被处理器执行时以实现一种地震数据静校正处理方法。该计算机设备的显示屏可以是液晶显示屏或者电子墨水显示屏，该计算机设备的输入装置可以是显示屏上覆盖的触摸层，也可以是计算机设备外壳上设置的按键、轨迹球或触控板，还可以是外接的键盘、触控板或鼠标等。In this embodiment, a computer device is provided. Its internal structure diagram can be shown in Figure 4. The computer equipment includes a processor, memory, a network interface, a display screen, and an input device connected by a system bus. Therein, the processor of the computing device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium, an internal memory. The nonvolatile storage medium stores an operating system and a computer program, and deploys a database configured to store a first-arrival pickup neural network model and a static correction processing neural network model. The internal memory provides an environment for the execution of the operating system and computer programs in the non-volatile storage medium. The network interface of the computer device is configured to communicate with other computer devices. The computer program, when executed by the processor, implements a static correction processing method for seismic data. The display screen of the computer equipment may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment may be a touch layer covered on the display screen, or a button, a trackball or a touchpad set on the shell of the computer equipment , or an external keyboard, trackpad, or mouse.

本领域技术人员可以理解，图4中示出的结构，仅仅是与本申请方案相关的部分结构的框图，并不构成对本申请方案所应被配置为其上的计算机设备的限定，具体的计算机设备可以包括比图中所示更多或更少的部件，或者组合某些部件，或者具有不同的部件布置。Those skilled in the art can understand that the structure shown in FIG. 4 is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer equipment on which the solution of the present application should be configured. A device may include more or fewer components than shown in the figures, or combine certain components, or have a different arrangement of components.

在一个实施例中，提供了一种计算机设备，包括存储器和处理器，该存储器存储有计算机程序，该处理器执行计算机程序时实现以下步骤：In one embodiment, a computer device is provided, including a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:

具体地，将需要进行初至拾取的地震数据提取出地震道数据体，按道顺序输入到初至拾取神经网络模型中。经过初至拾取神经网络模型计算后，输出层的节点值为1的节点对应的地震道的采样位置作为目标道的初至时间。对所有需要初至拾取的地震数据依次输入到神经网络中，最终获取所有数据的初至时间。Specifically, a seismic trace data volume is extracted from the seismic data that needs first-arrival picking, and is input into the first-arrival picking neural network model according to the track sequence. After the first arrival picking neural network model calculation, the sampling position of the seismic trace corresponding to the node whose node value of the output layer is 1 is taken as the first arrival time of the target trace. All the seismic data that needs to be picked up by the first arrival are input into the neural network in turn, and finally the first arrival time of all the data is obtained.

本实施例中，提取地震数据每一道的炮点坐标和高程(Sx，Sy，Sz)，检波点坐标和高程(Rx，Ry，Rz)，将提取的六个参数输入到神经网络的输入层，将输入道的初至时间作为输出层的验证数据。依次对目标地震数据的每一道进行神经网络训练。In this embodiment, the shot coordinates and elevations (Sx, Sy, Sz) and the receiver coordinates and elevations (Rx, Ry, Rz) of each track of the seismic data are extracted, and the extracted six parameters are input into the input layer of the neural network , the first arrival time of the input channel is used as the verification data of the output layer. Neural network training is performed on each track of the target seismic data in turn.

在一个实施例中，处理器执行计算机程序时还实现以下步骤：In one embodiment, the processor further implements the following steps when executing the computer program:

构建包括第一输入层、第一中间层和第一输出层的初至拾取神经网络，其中，所述第一输入层被配置为输入一道地震数据，通过所述第一中间层计算后，所述第一输出层输出一道与输入的震道数据的样点数一致的数据，输出的数据中每个样点值为第一样点值或第二样点值。Constructing a first-arrival picking neural network including a first input layer, a first intermediate layer and a first output layer, wherein the first input layer is configured to input a piece of seismic data, and after calculation by the first intermediate layer, the The first output layer outputs a piece of data that is consistent with the sample number of the input seismic trace data, and each sample point value in the output data is a first sample point value or a second sample point value.

具体地，首先，对初至拾取神经网络结构进行设置。按照一个输入层，两个中间层，一个输出层设计神经网络结构，输入层的节点数与需要初至拾取的地震数据采样点数一致，中间两层节点数和输出层节点数与输入层节点数一致。随后，生成训练样本数据。选取部分需要进行初至拾取的地震数据作为训练样本数据，首先对训练样本地震数据进行人工拾取初至，获取精确的初至时间。根据神经网络结构，提取样本地震数据的数据体，去除样本地震数据的文件头描述信息和每道的道头信息。验证数据体的获取是按照地震数据体生成同样大小的数据体，数据体中每个采样点值都设置为0，然后在每一道的初至时间对应的采样点位置的值设置为1，表示这一道数据的初至时间位置。然后，确定训练参数。神经网络的训练参数是确定训练效果的关键因素，考虑到训练的计算量和精度，通过循环次数和误差量两个参数控制神经网络训练。训练误差确定训练的精度，防止出现过拟合。循环次数控制训练的计算量，防止陷入多次循环，无法正常结束。Specifically, first, the first-arrival picking neural network structure is set. The neural network structure is designed according to one input layer, two middle layers, and one output layer. The number of nodes in the input layer is the same as the number of seismic data sampling points that need to be picked up first, and the number of nodes in the middle two layers and the number of nodes in the output layer are the same as the number of nodes in the input layer. Consistent. Subsequently, training sample data is generated. Part of the seismic data that needs to be picked for the first arrival is selected as the training sample data. First, the first arrival of the training sample seismic data is manually picked to obtain the accurate first arrival time. According to the neural network structure, the data body of the sample seismic data is extracted, and the file header description information of the sample seismic data and the track header information of each track are removed. The acquisition of the verification data volume is to generate a data volume of the same size according to the seismic data volume. The value of each sampling point in the data volume is set to 0, and then the value of the sampling point position corresponding to the first arrival time of each track is set to 1, indicating that The first arrival time position of this data. Then, the training parameters are determined. The training parameters of the neural network are the key factors to determine the training effect. Considering the calculation amount and accuracy of the training, the training of the neural network is controlled by two parameters: the number of cycles and the amount of error. The training error determines the accuracy of the training and prevents overfitting. The number of loops controls the amount of calculation for training, preventing it from falling into multiple loops and failing to end normally.

实施例7Example 7

本实施例中，提供了一种计算机可读存储介质，其上存储有计算机程序，计算机程序被处理器执行时实现以下步骤：In this embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

在一个实施例中，计算机程序被处理器执行时还实现以下步骤：In one embodiment, the computer program further implements the following steps when executed by the processor:

以上所述，仅为本申请的具体实施方式，但本申请的保护范围并不局限于此，任何熟悉本技术领域的技术人员在本申请揭露的技术范围内，可轻易想到变化或替换，都应涵盖在本申请的保护范围之内。因此，本申请的保护范围应以所述权利要求的保护范围为准。The above are only specific embodiments of the present application, but the protection scope of the present application is not limited to this. should be covered within the scope of protection of this application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

一种地震数据静校正处理方法，包括：A static correction processing method for seismic data, comprising:

获取目标地震数据；Obtain target seismic data;

将所述目标地震数据输入至预先训练的初至拾取神经网络模型进行训练，获得包含第一初至时间数据的初至拾取数据；Inputting the target seismic data into a pre-trained first-arrival picking neural network model for training to obtain first-arriving picking data including first first-arrival time data;

将包含所述第一初至时间数据的所述初至拾取数据输入至静校正处理神经网络进行训练，获得静校正处理后的初至拾取数据以及第二初至时间数据；Inputting the first-arrival pickup data including the first first-arrival time data into a static correction processing neural network for training to obtain first-arrival pickup data and second first-arrival time data after static correction processing;

根据静校正处理后的所述第二初至时间数据和所述第一初至时间数据，计算得到静校正量；Calculate the static correction amount according to the second first arrival time data and the first first arrival time data after the static correction processing;

根据所述静校正量对地震数据进行校正。The seismic data is corrected according to the static correction amount.
根据权利要求1所述的方法，其中，所述将所述目标地震数据输入至预先训练的初至拾取神经网络模型进行训练，获得包含初至时间的初至拾取数据的步骤之前包括：The method according to claim 1, wherein the step of inputting the target seismic data into a pre-trained first-arrival picking neural network model for training, before the step of obtaining the first-arriving picking data including the first-arrival time, comprises:

获取预设格式的样本数据，其中，所述预设格式的样本数据为炮集数据的每一道数据，地震道数据的格式包括文件头、每道的道头数据和道数据体，所述道数据体记录每个采样点上的振幅值；Obtain sample data in a preset format, wherein the sample data in the preset format is each trace of shot collection data, and the format of the seismic trace data includes a file header, trace header data for each trace, and a trace data body. The data volume records the amplitude value at each sampling point;

将所述预设格式的样本数据输入至初至拾取神经网络进行训练，得到所述初至拾取神经网络模型。The sample data in the preset format is input into the first-arrival picking neural network for training, and the first-arriving picking neural network model is obtained.
根据权利要求2所述的方法，其中，所述获取预设格式的样本数据的步骤之前还包括：The method according to claim 2, wherein before the step of acquiring the sample data in a preset format, the step further comprises:

构建包括第一输入层、第一中间层和第一输出层的初至拾取神经网络，其中，所述第一输入层被配置为输入一道地震数据，通过所述第一中间层计算后，所述第一输出层输出一道与输入的震道数据的样点数一致的数据，输出的数据中每个样点值为第一样点值或第二样点值。Constructing a first-arrival picking neural network including a first input layer, a first intermediate layer and a first output layer, wherein the first input layer is configured to input a piece of seismic data, and after calculation by the first intermediate layer, the The first output layer outputs a piece of data that is consistent with the sample number of the input seismic trace data, and each sample point value in the output data is a first sample point value or a second sample point value.
根据权利要求3所述的方法，其中，所述将所述目标地震数据输入至预先训练的初至拾取神经网络模型进行训练，获得包含第一初至时间数据的初至拾取数据的步骤包括：The method according to claim 3, wherein the step of inputting the target seismic data into a pre-trained first-arrival picking neural network model for training, and obtaining the first-arriving picking data including the first first-arriving time data comprises:

将所述目标地震数据输入至预先训练的初至拾取神经网络模型进行训练，所述初至拾取神经网络模型输出包含样点值的所述初至拾取数据，其中，所述样点值包括第一样点值和第二样点值；The target seismic data is input into a pre-trained first-arrival picking neural network model for training, and the first-arrival picking neural network model outputs the first-arrival picking data including sample point values, wherein the sample point values include The first sample value and the second sample value;

通过所述初至拾取神经网络模型提取出所述样点值为所述第一样点值的所述初至拾取数据，获取所述第一样点值对应的所述初至拾取数据的所述第一初至时间数据。The first-arrival picking data whose sample point value is the first sample point value is extracted through the first-arrival picking neural network model, and all the first-arriving picking data corresponding to the first sample point value are obtained. Describe the first arrival time data.
根据权利要求1所述的方法，其中，所述将包含所述第一初至时间数据的所述初至拾取数据输入至静校正处理神经网络进行训练，获得静校正处理后的初至拾取数据以及第二初至时间数据的步骤包括：The method according to claim 1, wherein the first-arrival pickup data including the first first-arrival time data is input into a static correction processing neural network for training to obtain the first-arrival pickup data after static correction processing and the second first arrival time data steps include:

将包含所述第一初至时间数据的所述初至拾取数据输入至静校正处理神经网络进行训练；inputting the first arrival pickup data including the first first arrival time data into a static correction processing neural network for training;

通过所述静校正处理神经网络，对需要静校正处理的目标地震数据重置静校正的高程，输出目标高程的初至时间。Through the static correction processing neural network, the static correction elevation is reset for the target seismic data that needs static correction processing, and the first arrival time of the target elevation is output.
根据权利要求1所述的方法，其中，所述将包含所述第一初至时间数据的所述初至拾取数据输入至静校正处理神经网络进行训练，获得静校正处理后的初至拾取数据以及第二初至时间数据的步骤之前还包括：The method according to claim 1, wherein the first-arrival pickup data including the first first-arrival time data is input into a static correction processing neural network for training, and the static correction processed first-arrival pickup data is obtained And the step before the second first arrival time data also includes:

构建包括第二输入层、第二中间层和第二输出层的静校正处理神经网络。A static correction processing neural network is constructed including a second input layer, a second intermediate layer, and a second output layer.
根据权利要求1-6任一项中所述的方法，其中，所述根据静校正处理后的所述第二初至时间数据和所述第一初至时间数据，计算得到静校正量的步骤包括：The method according to any one of claims 1-6, wherein the step of calculating the static correction amount according to the second first arrival time data and the first first arrival time data after static correction processing include:

计算静校正处理后的所述第二初至时间数据和所述第一初至时间数据之差，得到所述静校正量。The static correction amount is obtained by calculating the difference between the second first arrival time data and the first first arrival time data after the static correction process.
一种地震数据静校正处理装置，包括：A seismic data static correction processing device, comprising:

目标地震数据获取模块，被配置为获取目标地震数据；a target seismic data acquisition module, configured to acquire target seismic data;

第一神经网络训练模块，被配置为将所述目标地震数据输入至预先训练的初至拾取神经网络模型进行训练，获得包含第一初至时间数据的初至拾取数据；a first neural network training module, configured to input the target seismic data into a pre-trained first-arrival picking neural network model for training, and obtain first-arriving picking data including first first-arrival time data;

第二神经网络训练模块，被配置为将包含所述第一初至时间数据的所述初至拾取数据输入至静校正处理神经网络进行训练，获得静校正处理后的初至拾取数据以及第二初至时间数据；The second neural network training module is configured to input the first-arrival pickup data including the first first-arrival time data into a static correction processing neural network for training, and obtain the static-corrected first-arrival pickup data and the second first arrival time data;

静校正量计算模块，被配置为根据静校正处理后的所述第二初至时间数据和所述第一初至时间数据，计算得到静校正量；a static correction amount calculation module, configured to calculate a static correction amount according to the second first arrival time data and the first first arrival time data after the static correction processing;

静校正模块，被配置为根据所述静校正量对地震数据进行校正。A static correction module configured to correct the seismic data according to the static correction amount.
一种计算机设备，包括存储器和处理器，所述存储器存储有计算机程序，所述处理器执行所述计算机程序时实现权利要求1至7中任一项所述方法的步骤。A computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method of any one of claims 1 to 7 when the processor executes the computer program.
一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序被处理器执行时实现权利要求1至7中任一项所述的方法的步骤。A computer-readable storage medium having stored thereon a computer program that, when executed by a processor, implements the steps of the method of any one of claims 1 to 7.