Seismic exploration velocity analysis method and device, computer device and storage medium

Abstract

The application provides a seismic exploration velocity analysis method and device, computer equipment and a storage medium, the method comprising processing a depth domain velocity model set based on a two-way wave equation to obtain trace set data; processing the trace set data based on a stacking velocity spectrum production formula to obtain sample set data; extracting a time domain root mean square velocity curve from the depth domain velocity model set to obtain label set data; and training a preset deep learning model based on the sample set data and the label set data to obtain a seismic exploration velocity modeling system for seismic exploration velocity analysis. The calculated label set data and sample set data are input into the preset deep learning model to train the preset deep learning model, determine the parameters in the preset deep learning model, and obtain the seismic exploration velocity modeling system. In actual exploration, the stacking velocity spectrum data is input into the seismic exploration velocity modeling system, and the result of automatic interpretation of the stacking velocity spectrum can be calculated.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of seismic exploration technology, and particularly relates to a seismic exploration velocity analysis method and device, computer equipment and a storage medium. BACKGROUND

[0002] Under the theoretical framework of seismic exploration, seismic data processing can be summarized as an inversion problem, and the key of seismic exploration is the inversion of underground physical parameters. After obtaining the velocity model, seismic migration imaging and subsequent reservoir inversion and interpretation work are performed. The current seismic exploration velocity inversion mainly includes two categories of time domain and depth domain. The velocity analysis in the time domain is mainly based on picking up the maximum energy of the stacked velocity spectrum to obtain a seismic root mean square velocity model. Velocity modeling in the depth domain includes reflection, refraction tomography and full waveform inversion, which are the most commonly used techniques in velocity modeling.

[0003] However, in the velocity modeling process in the time domain of seismic exploration, a large number of manual interactions and time-consuming velocity analysis links are involved, resulting in a long analysis time of seismic exploration velocity. SUMMARY

[0004] Therefore, it is necessary to provide a seismic exploration velocity analysis method, device, computer equipment and storage medium to solve the above technical problems.

[0005] A seismic exploration velocity analysis method comprises the following steps:

[0006] processing a depth domain velocity model set based on a two-way wave equation to obtain gather data;

[0007] processing the gather data based on a stacked velocity spectrum formula to obtain sample set data;

[0008] extracting a time domain root mean square velocity curve from the depth domain velocity model set to obtain label set data;

[0009] training a preset depth learning model based on the sample set data and the label set data to obtain a seismic exploration velocity modeling system for seismic exploration velocity analysis.

[0010] In one embodiment, the step of processing a depth domain velocity model set based on a two-way wave equation to obtain gather data comprises the following steps:

[0011] obtaining the geological conditions of a plurality of seismic exploration work areas, analyzing the geological conditions of each seismic exploration work area to obtain geological structure features, and establishing the depth domain velocity model set based on the geological structure features.

[0012] In one of the embodiments, the step of processing the depth domain velocity model set based on the two-way wave equation to obtain the gather data comprises:

[0013] forward modeling the depth domain velocity model set based on the two-way wave equation to obtain the synthetic seismic data;

[0014] processing the synthetic seismic data to obtain the gather data.

[0015] In one of the embodiments, the step of processing the gather data based on the stacking velocity spectrum formulation to obtain the sample set data comprises:

[0016] extracting features from the gather data based on the stacking velocity spectrum formulation to obtain gather features, and converting the gather features into stacking velocity spectrum data;

[0017] processing the stacking velocity spectrum data to obtain the sample set data.

[0018] In one of the embodiments, the step of extracting the time domain RMS velocity curve from the depth domain velocity model set to obtain the label set data comprises:

[0019] extracting gather point velocities from the depth domain velocity model set to obtain a layer velocity curve;

[0020] converting the layer velocity curve into the time domain RMS velocity curve;

[0021] processing the time domain RMS velocity curve to obtain the label set data.

[0022] In one of the embodiments, the step of processing the time domain RMS velocity curve comprises:

[0023] processing the time domain RMS velocity curve based on maximum-minimum velocity normalization.

[0024] A seismic exploration velocity analysis device comprises:

[0025] a gather data extraction module configured to process a depth domain velocity model set based on a two-way wave equation to obtain gather data;

[0026] a sample set data acquisition module configured to process the gather data based on a stacking velocity spectrum formulation to obtain sample set data;

[0027] a label set data acquisition module configured to extract a time domain RMS velocity curve from the depth domain velocity model set to obtain label set data;

[0028] The model training module trains a preset deep learning model based on the sample set data and the label set data to obtain a seismic exploration velocity modeling system for seismic exploration velocity analysis.

[0029] In one of the embodiments, the seismic exploration velocity analysis device comprises:

[0030] The velocity model establishing module is configured to obtain geological conditions of a plurality of seismic exploration work areas, analyze the geological conditions of each of the seismic exploration work areas to obtain geological structure features, obtain the geological conditions of the plurality of seismic exploration work areas based on the geological structure features, analyze the geological conditions of each of the seismic exploration work areas to obtain the geological structure features, and establish the deep domain velocity model set based on the geological structure features.

[0031] A computer device comprises a memory and a processor, and the memory stores a computer program, wherein the processor implements the following steps when executing the computer program:

[0032] The deep domain velocity model set is processed based on a two-way wave equation to obtain trace set data.

[0033] The trace set data is processed based on a stacking velocity spectrum production formula to obtain sample set data.

[0034] A time domain root mean square velocity curve is extracted from the deep domain velocity model set to obtain label set data.

[0035] A preset deep learning model is trained based on the sample set data and the label set data to obtain a seismic exploration velocity modeling system for seismic exploration velocity analysis.

[0036] A computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the following steps:

[0037] The deep domain velocity model set is processed based on a two-way wave equation to obtain trace set data.

[0038] The trace set data is processed based on a stacking velocity spectrum production formula to obtain sample set data.

[0039] A time domain root mean square velocity curve is extracted from the deep domain velocity model set to obtain label set data.

[0040] A preset deep learning model is trained based on the sample set data and the label set data to obtain a seismic exploration velocity modeling system for seismic exploration velocity analysis.

[0041] The aforementioned seismic exploration velocity analysis method, apparatus, computer equipment, and storage medium input the calculated label set data and sample set data into a preset deep learning model, train the preset deep learning model, determine the parameters in the preset deep learning model, and obtain a seismic exploration velocity modeling system. Therefore, in actual exploration, the superimposed velocity spectrum data can be input into the seismic exploration velocity modeling system, and the results of automatically interpreting the superimposed velocity spectrum can be calculated. In the process of calculating the seismic exploration velocity analysis results, no manual interaction is required, shortening the analysis time for seismic exploration velocities. Attached Figure Description

[0042] Figure 1 This is a flowchart illustrating a seismic exploration velocity analysis method in one embodiment;

[0043] Figure 2 This is a structural block diagram of a seismic exploration velocity analysis device in one embodiment;

[0044] Figure 3 This is an internal structural diagram of a computer device in one embodiment;

[0045] Figure 4 This is a flowchart illustrating the seismic exploration velocity analysis method in another embodiment;

[0046] Figure 5 Here is an example diagram of a depth domain velocity model in one embodiment;

[0047] Figure 6 This is a shot gather seismic record simulated using the acoustic wave equation in one embodiment, along with an example image after gain.

[0048] Figure 7 This is an example diagram of CMP arrangement of seismic records for shot gather record sorting in one embodiment;

[0049] Figure 8 This is an example diagram of the stacked velocity spectrum prepared from CMP seismic records in one embodiment, and an example diagram after segmented normalization.

[0050] Figure 9 This is an example diagram illustrating the conversion of the velocity curve at the CMP point into tag data in one embodiment.

[0051] Figure 10 This is a network model diagram built in one embodiment;

[0052] Figure 11 This is a diagram showing the result of automatic interpretation of superimposed velocity spectrum using a neural network model in one embodiment. Detailed Implementation

[0053] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application is further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and not to limit the present application.

[0054] Embodiment one

[0055] In this embodiment, as shown in Figure 1 A seismic exploration velocity analysis method is provided, which comprises:

[0056] In step 110, a set of depth domain velocity models is processed based on a two-way wave equation to obtain gather data.

[0057] In this embodiment, a plurality of depth domain velocity models are first acquired to establish a set of depth domain velocity models. Specifically, according to the geological conditions of a seismic exploration work area, specific geological structure features are analyzed and abstractly represented to obtain a plurality of depth domain velocity models, and a set of depth domain velocity models with different geological features is established, so as to ensure the broadness of the model covering the geological structure features of the actual processing work area. Figure 5 The set of depth domain velocity models.

[0058] In this embodiment, the set of depth domain velocity models contains a plurality of depth domain velocity models, and each depth domain velocity model is processed by a two-way wave equation to obtain gather data.

[0059] In step 120, the gather data is processed based on a stacking velocity spectrum production formula to obtain sample set data.

[0060] In this embodiment, the stacking velocity spectrum data is acquired from the gather data as sample set data by the stacking velocity spectrum production formula. The corresponding gather data is generated by using a sound wave forward simulation method, and then the stacking velocity spectrum data is generated by scanning stacking as sample set data.

[0061] In step 130, a time domain root mean square velocity curve is extracted from the set of depth domain velocity models to obtain label set data.

[0062] The time domain root mean square velocity curve is used because the time domain root mean square velocity curve has a more direct spatial corresponding relationship with the stacking velocity spectrum data, which is beneficial to extract the corresponding local features. Moreover, the time domain root mean square velocity curve and the stacking velocity spectrum data have the same time length, which is easier to determine the specific depth.

[0063] In this embodiment, since the stacking velocity spectrum data is related to time, in order to associate the sample set data and the label set data, the time domain root mean square velocity curve is extracted from the set of depth domain velocity models, wherein the time domain root mean square velocity curve is related to time.

[0064] In step 140, a preset deep learning model is trained based on the sample set data and the label set data, and a seismic exploration velocity modeling system for seismic exploration velocity analysis is obtained.

[0065] In this embodiment, the sample set data and the label set data are input into the preset deep learning model, the preset deep learning model is trained, the parameters of the preset deep learning model are determined, and the trained preset deep learning model is obtained as the seismic exploration velocity modeling system.

[0066] In the above embodiment, the seismic exploration velocity analysis method of the present application inputs the calculated label set data and sample set data into the preset deep learning model, trains the preset deep learning model, determines the parameters in the preset deep learning model, and obtains the seismic exploration velocity modeling system. Thus, in actual exploration, the stacked velocity spectrum data can be input into the seismic exploration velocity modeling system, and the result of automatic interpretation of the stacked velocity spectrum can be calculated.

[0067] In one embodiment, the step of training the preset deep learning model based on the sample set data and the label set data to obtain the seismic exploration velocity modeling system for seismic exploration velocity analysis further comprises:

[0068] The measured stacked velocity spectrum data is obtained, the measured stacked velocity spectrum data is input into the seismic exploration velocity modeling system, the measured stacked velocity spectrum data is processed by the seismic exploration velocity modeling system, and interpretation result data is obtained.

[0069] In this embodiment, the measured stacked velocity spectrum data is the stacked velocity spectrum data obtained in actual exploration and needing to be tested, the seismic exploration velocity modeling system calculates the measured stacked velocity spectrum data, and obtains the seismic exploration velocity analysis result. In the process of calculating the seismic exploration velocity analysis result, no manual interaction is required, and the analysis time of the seismic exploration velocity is shortened.

[0070] In one embodiment, the step of processing the depth domain velocity model set based on the two-way wave equation to obtain the gather data comprises: obtaining the geological conditions of a plurality of seismic exploration work areas, analyzing the geological conditions of each of the seismic exploration work areas to obtain geological structure features, and establishing the depth domain velocity model set based on the geological structure features.

[0071] In the embodiment, in order to obtain more sample set data and label set data for training the preset deep learning model, multiple deep velocity domain velocity models are established, so that more sample set data and label set data can be obtained from the multiple deep velocity domain velocity models as a training set, a large amount of training set is provided, and parameters in the preset deep learning model are better calculated, the seismic exploration velocity modeling system is more accurate in analyzing stacked velocity spectrum data, and more accurate calculation results are output.

[0072] In the embodiment, according to the geological complexity of the domestic seismic exploration work area, the specific geological structure characteristics are analyzed and abstractly represented, and a deep domain velocity model set with different geological characteristics is established, so as to ensure the broadness of the deep domain velocity model set covering the geological structure characteristics of the actual processing work area.

[0073] In the embodiment, the deep domain velocity model set is defined according to formula (1):

[0074] V depth = Γ(V m , N dip , M layer , D abnormal )

[0075] V m ∈[1000,7000],N dip ∈[-35°,35°]and different with layers,

[0076] M laver ∈[1,14],D abnormal ∈[0,1]

[0077] Wherein, V depth represents the total velocity parameter of the current deep domain velocity model, the function Γ is an abstract velocity model representation, V m is the layer velocity of the Nth layer of the current deep domain velocity model, wherein N is between 0-M layer , Mlayer is the number of layers, the dip angle of the layer is N dip , wherein N dip is between-35 and 35 degrees, and D abnormal represents whether there is an abnormal body in the layer velocity. That is, according to the number of layers Mlayer, the dip angle Ndip, the velocity parameter Vm, the depth parameter Dabnormal of the generated velocity abnormal body, and the generated deep domain velocity model. Based on the dip angle, the velocity range, the geological abnormal body and the number of layers, the deep domain velocity model is established, so that the deep domain velocity model set is more comprehensive and adapts to the geological complexity of the domestic seismic exploration work area.

[0078] In one embodiment, the step of processing the depth domain velocity model set based on the two-way wave equation to obtain the gather data comprises:

[0079] Step 111, performing two-way wave equation forward modeling on the depth domain velocity model set to obtain synthetic seismic data.

[0080] Step 112, performing gain compensation processing on the synthetic seismic data to obtain the gather data.

[0081] In this embodiment, the two-way wave equation is an acoustic two-way wave equation, and the depth domain velocity model set is processed by the acoustic two-way wave equation to prepare synthetic seismic data through forward modeling strategy.

[0082] In this embodiment, the step of performing gain compensation processing on the synthetic seismic data to obtain the gather data comprises: performing gain compensation processing on the synthetic seismic data to obtain compensated seismic data; and sorting the compensated seismic data in a CMP (Common Middle Point) gather arrangement format to obtain the gather data. It should be understood that when the synthetic seismic data is prepared into stacked velocity spectrum data, the input data is sorted in the CMP arrangement format.

[0083] In one embodiment, the step of processing the gather data based on the stacked velocity spectrum production formula to obtain the sample set data comprises:

[0084] Step 121, extracting features from the gather data based on the stacked velocity spectrum production formula to obtain gather features, and converting the gather features into stacked velocity spectrum data;

[0085] Step 122, performing segmented normalization processing on the stacked velocity spectrum data to obtain the sample set data.

[0086] In this embodiment, the gather data is extracted according to formula (2):

[0087]

[0088] wherein Spec amp represents the prepared stacked velocity spectrum data, v represents a current scanning velocity, the range of v is between v start and v end , the specific value of v is determined according to the velocity range of the depth domain velocity model, dt is a data sampling point interval, ntrace represents the total number of seismic data, wherein CMP represents the gather data to be processed, nt is the number of sampling points, it is the index of the sampling point, ntrace is the total number of gather data, itrace is the index of the number of traces, v start is the start of the scanning velocity, and v endFor the end point of the scanning speed, the speed parameter dt is the sampling interval, v is the current scanning speed, X offset is the offset length corresponding to the current seismic trace. In this embodiment, the velocity range of the depth domain velocity model is 1000m / s-7350m / s. In this embodiment, the data sampling point interval dt is 0.5ms.

[0089] In this embodiment, the stacked velocity spectrum data is segmented and normalized according to formula (3) to obtain sample set data;

[0090] Spec norm (v, it) = Spec amp (v, it) / Max(Spec amp (v, it))it∈[t_start, t_end] (3)

[0091] wherein Spec norm represents sample set data, and t_start and t_end depend on the time window size. In this embodiment, the time window length is 10 sampling points.

[0092] In one embodiment, the step of extracting a time domain root mean square velocity curve from the depth domain velocity model set to obtain label set data comprises:

[0093] Step 131, extracting trace set point velocity from the depth domain velocity model set to obtain a layer velocity curve;

[0094] Step 132, converting the layer velocity curve into the time domain root mean square velocity curve;

[0095] Step 133, normalizing the time domain root mean square velocity curve to obtain label set data.

[0096] In this embodiment, the CMP center point layer velocity curve of the depth domain velocity model is extracted to obtain a layer velocity curve.

[0097] In this embodiment, the layer velocity curve is converted into a time domain root mean square velocity curve based on the DIX formula.

[0098] In one embodiment, the step of normalizing the time domain root mean square velocity curve comprises:

[0099] The time domain root mean square velocity curve is normalized based on the maximum and minimum velocity.

[0100] In this embodiment, the time domain root mean square velocity curve is normalized based on the maximum and minimum velocity according to formula (4):

[0101] V = (V-V min ) / (Vmax -V min )

[0102] wherein V represents a seismic velocity model, V min and V max represent a set normalized velocity range, V min represents a normalized minimum velocity, V max represents a normalized maximum velocity. In this embodiment, the normalized velocity range is 1000 m / s-7350 m / s, V min is 1000 m / s, and V max is 7350 m / s.

[0103] In this embodiment, in the maximum and minimum velocity normalization process, the maximum and minimum velocities are determined according to the characteristics of the work area.

[0104] In step 140, the preset deep learning model is pre-built, but the parameters of the preset deep learning model are not determined. The sample set data obtained in step 120 and the label set data obtained in step 130 are input into the preset deep learning model, the parameters of the preset deep learning model are determined, and the parameters of the trained preset deep learning model are obtained as a seismic exploration velocity modeling system. The parameters of the preset deep learning model are saved to a local disk, and the preset deep learning model saved to the local disk is deployed in a distributed processing cluster, so that different users can use the same preset deep learning model to interpret the stacked velocity spectrum respectively, improving the processing efficiency. When the parameters of the preset deep learning model are deployed in the distributed processing cluster, each cluster node user uses the preset deep learning model to first load the parameters of the preset deep learning model into the preset deep learning model, and then use the preset deep learning model to automatically interpret the stacked velocity spectrum to be processed, improving the work efficiency.

[0105] In an embodiment, the preset deep learning model is a neural network model based on deep learning.

[0106] In this embodiment, the neural network model is a network structure based on Unet+++.

[0107] It should be understood that, although each step in the flowchart of Figure 1 is shown in sequence according to the direction of the arrow, these steps are not necessarily executed in sequence according to the direction of the arrow. Unless otherwise specified herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other orders. Moreover, Figure 1At least one part of the steps in the method can include multiple sub-steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution order of the sub-steps or stages is not necessarily sequential, but can be executed alternately or alternately with at least one part of other steps or sub-steps or stages of other steps.

[0108] Embodiment two

[0109] In this embodiment, a seismic exploration velocity analysis method is provided, comprising:

[0110] In step 410, the geological conditions of a plurality of seismic exploration work areas are obtained, the geological structure characteristics of each seismic exploration work area are analyzed, and a set of depth domain velocity models is established based on the geological structure characteristics.

[0111] In this embodiment, according to the complex geological conditions of domestic seismic exploration work areas, specific geological structure characteristics are analyzed and abstractly represented, and a set of depth domain velocity models with different geological characteristics is established to ensure the broadness of the model covering the geological structure characteristics of the actual processing work area.

[0112] In this embodiment, by analyzing the geological conditions of each seismic exploration work area in China, the following geological structure characteristics are summarized, and the main influencing factors are dip angle, velocity range, geological anomaly body and number of strata. In this embodiment, the following formula is used to define the set of depth domain velocity models:

[0113] V depth = Γ(V m , N dip , M layer , D abnormal )

[0114] V m ∈[1000, 7000], N dip ∈|-35°, 35°|and different with layers,

[0115] M layer ∈[1, 14], D abnormal ∈[0, 1]

[0116] Wherein, V depth is the total velocity parameter of the current depth domain velocity model, the function Γ is the abstracted depth domain velocity model representation, V m is the layer velocity of the Nth layer of the current depth domain velocity model, wherein N is between 0-M layer , and the layer dip angle is N dip , wherein N dip is between -35 and 35 degrees, and Dabnormal indicates whether there is an abnormal body in the interval velocity. The depth domain velocity model is randomly generated according to the number of layers Mlayer, the dip angle Ndip, the velocity parameter Vm, and the depth parameter Dabnormal of the generated velocity abnormal body.

[0117] As shown in a model set graph of the depth domain velocity model, the depth domain velocity model set includes a plurality of depth domain velocity models, and different depth domain velocity models correspond to different geological conditions of the seismic exploration area. Figure 5

[0118] In step 420, the depth domain velocity model set is processed based on the two-way wave equation to obtain gather data.

[0119] In this embodiment, a large amount of synthetic seismic data is obtained by performing two-way wave equation forward modeling on the established depth domain velocity model, and the synthetic data obtained by modeling is gain compensated and extracted as CMP gathers.

[0120] In this embodiment, the synthetic data is prepared by using acoustic two-way wave forward modeling strategy. First, according to a large number of depth domain velocity models established in step 410, a unified observation system is laid out for all the depth domain velocity models in the following manner: the shot-receiver points are all placed on the ground surface, the shot points are located in the center of the model, and the receivers are equally spaced horizontally on the ground surface to ensure that the receivers on both sides of the shot point are the same distance apart; the obtained synthetic record data is automatically gain compensated and sorted into CMP arrangement format seismic data. Figure 6 It is an example graph of shot gather seismic records simulated by using acoustic wave equation. Figure 7 It is an example graph of CMP arrangement seismic records sorted from shot gather records.

[0121] In step 430, the gather data is processed based on a stacking velocity spectrum production formula to obtain sample set data.

[0122] In this embodiment, according to the stacking velocity spectrum production formula used in the conventional seismic exploration field, the CMP features are extracted and converted into stacking velocity spectrum data; the obtained stacking velocity spectrum data is converted into sample set data that can be input into a neural network by using a segmented normalization data enhancement algorithm.

[0123] In this embodiment, the gather data is feature extracted, mainly by using the following formula:

[0124]

[0125] wherein Spec amp indicates the prepared stacking velocity spectrum data, v indicates the current scanning velocity size, and the range of v is v start and v​end between them, the specific value of v is determined according to the model velocity range, in this embodiment, 1000m / s-7350m / s is adopted, dt is the data sampling point interval, the data sampling point interval dt adopts 0.5ms, ntrace represents the total number of seismic data, wherein CMP represents the seismic data to be processed currently. nt is the sampling point number, it is the index of the sampling point, ntrace is the total number of the cmp gather, itrace is the index of the trace, vstart and vend are the starting and ending velocity parameters of the scanning velocity, X offset is the offset length corresponding to the current trace.

[0126] Figure 8 is the offset length corresponding to the current trace.

[0127] In this embodiment, the stacked velocity spectrum data obtained by the above operation also needs to be segmented and normalized, and the processing operation is represented by the following formula:

[0128] Spec norm (v, it) = Spec amp (v, it) / Max (Spec amp (v, it)) it [t_start, t_end]

[0129] Where t_start and t_end depend on the time window size, Spec norm represents the sample set data. In this embodiment, the time window length adopted is 10 sampling points.

[0130] Step 440, extracting the time domain root mean square velocity curve from the depth domain velocity model set to obtain the label set data.

[0131] In this embodiment, the depth domain layer velocity curve at the CMP point is extracted from the depth domain velocity model set established in step 410, the depth domain layer velocity curve is converted into time domain RMS velocity curve data, and the maximum and minimum velocity range suitable for the work area characteristics is selected for normalization processing to obtain the label set data that can be input into the neural network. For the velocity spectrum data prepared in step 430, the depth domain layer velocity curve at the cmp point corresponding to the data is extracted.

[0132] In this embodiment, the depth domain velocity model obtained in step 410 is subjected to CMP center point layer velocity curve extraction, the DIX formula is adopted to convert the layer velocity curve into RMS velocity curve, and then the maximum and minimum velocity is adopted for normalization processing:

[0133] V = (V-V min ) / (V max -V min )

[0134] wherein, V represents a seismic velocity model, V min and V max represent a set of normalized velocity range. In this embodiment, the velocity range is normalized by 1000m / s-7350m / s. Figure 9 The label set data obtained after normalization is a label set instance graph, that is, a time domain root mean square velocity curve instance graph.

[0135] Step 450, training a preset deep learning model based on the sample set data and the label set data to obtain a seismic exploration velocity modeling system for seismic exploration velocity analysis.

[0136] In this embodiment, a neural network model of seismic exploration velocity analysis technology based on deep learning is built, and the sample set data prepared in step 430 and the label set data prepared in step 440 are input into the neural network model based on deep learning for training processing to obtain parameters of the neural network model, and the parameters of the neural network model are saved to a local disk. As shown in FIG. 5, a built network model. Figure 10 Figure 11 The seismic exploration velocity modeling system automatically interprets a result graph of the stacked velocity spectrum.

[0137] In this embodiment, a network structure based on Unet+++ is used. The sample set data and the label set data are input into the built neural network model for training, and after the training is completed, the parameters of the neural network model are saved to the local disk. The network model saved to the local disk is deployed in a distributed processing cluster, so that different users can use the same network model to interpret the stacked velocity spectrum respectively, and the processing efficiency is improved. In this embodiment, the obtained parameters of the neural network model are deployed in the distributed processing cluster, and when each cluster node user uses the module, the parameters of the neural network model are first loaded into the neural network, and then the neural network is used to automatically interpret the stacked velocity spectrum to be processed, thereby improving the work efficiency.

[0138] Embodiment Three

[0139] In this embodiment, as shown in FIG. 6, a neural network model of seismic exploration velocity analysis technology based on deep learning is built, and the sample set data prepared in step 430 and the label set data prepared in step 440 are input into the neural network model based on deep learning for training processing to obtain parameters of the neural network model, and the parameters of the neural network model are saved to a local disk. Figure 4 ​As shown in the first aspect, a seismic exploration velocity analysis method is provided, comprising: randomly generating a depth domain velocity model training set; performing acoustic wave equation forward modeling on the depth domain velocity model set to obtain synthetic seismic data, automatically performing gain compensation on the synthetic seismic data, and extracting CMP gather data, and preparing a stacking velocity spectrum through a stacking velocity spectrum preparation formula; segmenting and normalizing the stacking velocity spectrum into sample data; extracting CMP point velocities from the depth domain velocity model to obtain a layer velocity curve; converting the layer velocity curve into an RMS curve through a DIX formula; normalizing the RMS curve to obtain label data; inputting the sample data and the label data into a built convolutional neural network and training to obtain a trained neural network as a seismic exploration velocity modeling system. In actual seismic exploration, the stacking velocity spectrum data is input into the seismic exploration velocity modeling system, and the seismic exploration velocity modeling system outputs a prediction result.

[0140] Embodiment four

[0141] In this embodiment, as shown in the first aspect, Figure 2 A seismic exploration velocity analysis device is provided, comprising:

[0142] The gather data extraction module 210 processes the depth domain velocity model set based on the two-way wave equation to obtain gather data.

[0143] The sample set data acquisition module 220 processes the gather data based on the stacking velocity spectrum preparation formula to obtain sample set data.

[0144] The label set data acquisition module 230 extracts a time domain root mean square velocity curve from the depth domain velocity model set to obtain label set data.

[0145] The model training module 240 trains a preset deep learning model based on the sample set data and the label set data to obtain a seismic exploration velocity modeling system for seismic exploration velocity analysis.

[0146] In summary, the seismic exploration velocity analysis device of the present application inputs the calculated label set data and sample set data into a preset deep learning model, trains the preset deep learning model, determines the parameters in the preset deep learning model, and obtains a seismic exploration velocity modeling system. Thus, in actual exploration, the stacking velocity spectrum data can be input into the seismic exploration velocity modeling system, and the result of automatic interpretation of the stacking velocity spectrum can be calculated.

[0147] The gather data extraction module 210 processes the depth domain velocity model set based on the two-way wave equation to obtain gather data.

[0148] In this embodiment, the depth domain velocity model set contains multiple depth domain velocity models, and each depth domain velocity model is processed by the two-way wave equation to obtain gather data.

[0149] The sample set data acquisition module 220 processes the gather data based on the stacked velocity spectrum production formula to obtain sample set data.

[0150] In this embodiment, the stacked velocity spectrum data is obtained from the gather data by the stacked velocity spectrum production formula as the sample set data.

[0151] The label set data acquisition module 230 extracts the time domain root mean square velocity curve from the depth domain velocity model set to obtain label set data.

[0152] In this embodiment, since the stacked velocity spectrum data is related to time, in order to associate the sample set data and the label set data, the time domain root mean square velocity curve is extracted from the depth domain velocity model set, wherein the time domain root mean square velocity curve is related to time.

[0153] The model training module 240 trains a preset deep learning model based on the sample set data and the label set data to obtain a seismic exploration velocity modeling system for seismic exploration velocity analysis.

[0154] In this embodiment, the sample set data and the label set data are input into the preset deep learning model, the preset deep learning model is trained, the parameters of the preset deep learning model are determined, and the trained preset deep learning model is obtained as the seismic exploration velocity modeling system.

[0155] In the model training module 240, the preset deep learning model is pre-built, but the parameters of the preset deep learning model are not determined. The sample set data obtained by the sample set data acquisition module 220 and the label set data obtained by the label set data acquisition module 230 are input into the preset deep learning model, the parameters of the preset deep learning model are determined, and the parameters of the trained preset deep learning model are obtained as the seismic exploration velocity modeling system. The parameters of the preset deep learning model are saved to the local disk, and the preset deep learning model saved to the local disk is deployed in the distributed processing cluster, so that different users can use the same preset deep learning model to interpret the stacked velocity spectrum respectively, improving the processing efficiency. When the parameters of the preset deep learning model are deployed in the distributed processing cluster, each cluster node user uses the preset deep learning model to first load the parameters of the preset deep learning model into the preset deep learning model, and then uses the preset deep learning model to automatically interpret the stacked velocity spectrum to be processed, improving the work efficiency.

[0156] In one embodiment, the seismic exploration velocity analysis device comprises:

[0157] The velocity model establishing module 250 is configured to obtain geological conditions of multiple seismic exploration work areas, analyze the geological conditions of each of the seismic exploration work areas to obtain geological structure features, and establish the set of depth domain velocity models based on the geological structure features.

[0158] In the embodiment, the preset deep learning model is trained to obtain more sample set data and label set data, and multiple deep velocity domain velocity models are established. Thus, more sample set data and label set data can be obtained from the multiple deep velocity domain velocity models as a training set, a large amount of training set is provided, and the parameters in the preset deep learning model are better calculated. The seismic exploration velocity modeling system is more accurate in analyzing the stacked velocity spectrum data, and outputs more accurate calculation results.

[0159] In the embodiment, according to the geological complexity of the domestic seismic exploration work area, the specific geological structure features are analyzed and abstractly represented, and the set of depth domain velocity models with different geological features is established to ensure the broadness of the set of depth domain velocity models covering the geological structure features of the actual processing work area.

[0160] In the embodiment, the set of depth domain velocity models is defined according to formula (1):

[0161] V depth = Γ(V m , N dip , M layer , D abnormal )

[0162] V m ∈[1000,7000],N dip ∈[-35°,35°]and different with layers,

[0163] M layer ∈[1,14],D abnormal ∈[0,1]

[0164] wherein, V depth represents the total velocity parameter of the current depth domain velocity model, the function Γ is an abstract velocity model representation, V m represents the layer velocity of the Nth layer of the current depth domain velocity model, wherein N is between 0 and M layer , Mlayer represents the number of layers, and the layer dip angle of the layer is N dip , wherein N dip is between -35 and 35 degrees, and D abnormalIndicates whether there is an abnormal body in the layer velocity. That is, according to the number of layers Mlayer, the dip angle Ndip, the velocity parameter Vm, the depth parameter Dabnormal of the generated velocity abnormal body, and the generated depth domain velocity model. Based on the dip angle, the velocity range, the geological abnormal body, and the number of layers, the depth domain velocity model is established, so that the depth domain velocity model set is more comprehensive, and adapts to the geological complex situation of the domestic seismic exploration work area.

[0165] In one embodiment, the gather data extraction module 210 includes:

[0166] The seismic data synthesis unit 211 performs two-way wave equation forward modeling on the depth domain velocity model set to obtain synthetic seismic data.

[0167] The gather data acquisition unit 212 performs gain compensation processing on the synthetic seismic data to obtain gather data.

[0168] In this embodiment, the two-way wave equation is an acoustic two-way wave equation, and the depth domain velocity model set is processed by the acoustic two-way wave equation to prepare synthetic seismic data by forward modeling strategy.

[0169] In this embodiment, the step of performing gain compensation processing on the synthetic seismic data to obtain gather data includes: performing gain compensation processing on the synthetic seismic data to obtain compensated seismic data; and sorting the compensated seismic data in CMP arrangement format to obtain gather data. It should be understood that when preparing stack velocity spectrum data from synthetic seismic data, the input data is sorted in CMP format.

[0170] In one embodiment, the sample set data acquisition module 220 includes:

[0171] The velocity spectrum acquisition module 221 extracts features from the gather data based on the stack velocity spectrum formula to obtain gather features, and converts the gather features into stack velocity spectrum data.

[0172] The sample set acquisition module 222 performs segmented normalization processing on the stack velocity spectrum data to obtain sample set data.

[0173] In this embodiment, the gather data is characterized according to formula (2):

[0174]

[0175]

[0176] wherein Spec amp represents the prepared stack velocity spectrum data, v represents the current scanning velocity, and the range of v is v start and vend between the start and end points of the scan, v is the current scan velocity, dt is the sampling interval, and X start is the start of the scan velocity, v end is the end of the scan velocity, the velocity parameter dt is the sampling interval, v is the current scan velocity, and X offset is the offset length corresponding to the current seismic trace. In this embodiment, the velocity range of the depth domain velocity model is 1000 m / s-7350 m / s. In this embodiment, the data sampling point interval dt is 0.5 ms.

[0177] In this embodiment, the stacking velocity spectrum data is segmented and normalized according to formula (3) to obtain sample set data.

[0178] Spec norm (v, it) = Spec amp (v, it) / Max(Spec amp (v, it)) it e [t start, t end] (3)

[0179] where Spec norm represents the sample set data, and t start and t end depend on the time window size. In this embodiment, the time window length is 10 sampling points.

[0180] In one embodiment, the label set data acquisition module 230 includes:

[0181] The velocity extraction module 231 extracts the trace point velocity from the depth domain velocity model set to obtain a layer velocity curve.

[0182] The velocity conversion module 232 converts the layer velocity curve into a time domain root mean square velocity curve.

[0183] The label set acquisition module 233 normalizes the time domain root mean square velocity curve to obtain label set data.

[0184] In this embodiment, the CMP center point layer velocity curve is extracted from the depth domain velocity model.

[0185] In this embodiment, the layer velocity curve is converted into a time domain root mean square velocity curve based on the DIX formula.

[0186] In one embodiment, the normalization processing of the time domain root mean square velocity curve includes:

[0187] normalizing the time-domain root-mean-square velocity curve based on maximum and minimum velocities.

[0188] In this embodiment, the time-domain root-mean-square velocity curve is normalized based on maximum and minimum velocities according to formula (4):

[0189] V = (V-V min ) / (V max -V min )

[0190] wherein V represents a seismic velocity model, V min and V max represent a set normalized velocity range, V min represents a normalized minimum velocity, and V max represents a normalized maximum velocity. In this embodiment, the normalized velocity range is 1000 m / s-7350 m / s, V min is 1000 m / s, and V max is 7350 m / s.

[0191] The specific limitations of the seismic exploration velocity analysis device can be seen in the limitations of the seismic exploration velocity analysis method above, and will not be repeated here. Each unit in the above seismic exploration velocity analysis device can be realized based on software, hardware, and combinations thereof, in whole or in part. The above units can be embedded in or independent of the processor in the computer device in hardware form, or can be stored in the memory in the computer device in software form, so as to be called and executed by the processor to perform the operations corresponding to each unit.

[0192] Embodiment Five

[0193] In this embodiment, a computer device is provided. Its internal structure diagram can be as shown in Figure 3The computer device shown in the figure includes a processor, a memory, a network interface, a display screen and an input device connected based on a system bus. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program, and the non-volatile storage medium is deployed with a database configured to store data of geological conditions of a seismic exploration area and a depth domain velocity model. The internal memory provides an environment for running the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is configured to communicate with other computer devices deployed with application software. The computer program is executed by the processor to implement a seismic exploration velocity analysis method. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer overlaid on the display screen, or a button, trackball or touchpad arranged on the shell of the computer device, or an external keyboard, touchpad or mouse, etc.

[0194] Those skilled in the art can understand that, Figure 3 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement.

[0195] In one embodiment, a computer device is provided, including a memory storing a computer program and a processor configured to execute the computer program to implement the following steps:

[0196] In step 110, a set of depth domain velocity models is processed based on a two-way wave equation to obtain gather data.

[0197] In this embodiment, the set of depth domain velocity models includes a plurality of depth domain velocity models, and each depth domain velocity model is processed by the two-way wave equation to obtain gather data.

[0198] In step 120, the gather data is processed based on a stacking velocity spectrum production formula to obtain sample set data.

[0199] In this embodiment, the stacking velocity spectrum data is obtained from the gather data as the sample set data by the stacking velocity spectrum production formula.

[0200] In step 130, a time domain root mean square velocity curve is extracted from the set of depth domain velocity models to obtain label set data.

[0201] In the embodiment, since the stacked velocity spectrum data is related to time, in order to associate the sample set data and the label set data, the time domain root mean square velocity curve related to time is extracted from the depth domain velocity model set.

[0202] In step 140, the preset deep learning model is trained based on the sample set data and the label set data, and a seismic exploration velocity modeling system for seismic exploration velocity analysis is obtained.

[0203] In the embodiment, the sample set data and the label set data are input into the preset deep learning model, the preset deep learning model is trained, the parameters of the preset deep learning model are determined, and the trained preset deep learning model is obtained as the seismic exploration velocity modeling system.

[0204] In one embodiment, the processor, when executing the computer program, also implements the following steps:

[0205] The geological conditions of a plurality of seismic exploration work areas are obtained, the geological conditions of each of the seismic exploration work areas are analyzed to obtain geological structure features, and the depth domain velocity model set is established based on the geological structure features.

[0206] In one embodiment, the processor, when executing the computer program, also implements the following steps:

[0207] In step 111, two-way wave equation forward modeling is performed on the depth domain velocity model set to obtain synthetic seismic data.

[0208] In step 112, gain compensation processing is performed on the synthetic seismic data to obtain trace gather data.

[0209] In one embodiment, the processor, when executing the computer program, also implements the following steps:

[0210] In step 121, based on the stacked velocity spectrum production formula, the features are extracted from the trace gather data to obtain trace gather features, and the trace gather features are converted into stacked velocity spectrum data.

[0211] In step 122, the stacked velocity spectrum data is subjected to segmented normalization processing to obtain sample set data.

[0212] In one embodiment, the processor, when executing the computer program, also implements the following steps:

[0213] In step 131, the trace gather point velocity is extracted from the depth domain velocity model set to obtain a layer velocity curve.

[0214] In step 132, the layer velocity curve is converted into the time domain root mean square velocity curve.

[0215] Step 133, normalizing the time-domain root-mean-square velocity curve to obtain label set data.

[0216] Embodiment six

[0217] In this embodiment, a computer readable storage medium is provided, and the computer readable storage medium stores a computer program. The computer program is executed by a processor to implement the following steps:

[0218] Step 110, processing the depth-domain velocity model set based on the two-way wave equation to obtain gather data.

[0219] In this embodiment, the depth-domain velocity model set includes a plurality of depth-domain velocity models. Each depth-domain velocity model is processed by the two-way wave equation to obtain gather data.

[0220] Step 120, processing the gather data based on a stacking velocity spectrum production formula to obtain sample set data.

[0221] In this embodiment, the stacking velocity spectrum data is obtained from the gather data as the sample set data by the stacking velocity spectrum production formula.

[0222] Step 130, extracting a time-domain root-mean-square velocity curve from the depth-domain velocity model set to obtain label set data.

[0223] In this embodiment, since the stacking velocity spectrum data is related to time, in order to associate the sample set data and the label set data, the time-domain root-mean-square velocity curve is extracted from the depth-domain velocity model set, and the time-domain root-mean-square velocity curve is related to time.

[0224] Step 140, training a preset deep learning model based on the sample set data and the label set data to obtain a seismic exploration velocity modeling system for seismic exploration velocity analysis.

[0225] In this embodiment, the sample set data and the label set data are input into the preset deep learning model, the preset deep learning model is trained, the parameters of the preset deep learning model are determined, and the trained preset deep learning model is obtained as the seismic exploration velocity modeling system.

[0226] In summary, in this embodiment, the calculated label set data and sample set data are input into the preset deep learning model, the preset deep learning model is trained, the parameters in the preset deep learning model are determined, and the seismic exploration velocity modeling system is obtained. Therefore, in actual exploration, the stacking velocity spectrum data can be input into the seismic exploration velocity modeling system, and the result of automatic interpretation of the stacking velocity spectrum can be calculated.

[0227] In one embodiment, the computer program is executed by the processor to further implement the following steps:

[0228] Obtaining geological conditions of a plurality of seismic exploration work areas, analyzing the geological conditions of each of the seismic exploration work areas to obtain geological structure characteristics, and establishing the deep domain velocity model set based on the geological structure characteristics.

[0229] In the embodiment, in order to obtain more sample set data and label set data for training the preset deep learning model, a plurality of deep domain velocity models are established. Thus, more sample set data and label set data can be obtained from the plurality of deep domain velocity models as a training set, a large amount of training set is provided, and the parameters in the preset deep learning model are better calculated. The seismic exploration velocity modeling system is more accurate in analyzing the stacked velocity spectrum data, and outputs more accurate calculation results.

[0230] In the embodiment, according to the geological complexity of the domestic seismic exploration work area, the specific geological structure characteristics are analyzed and abstractly represented, and the deep domain velocity model set with different geological characteristics is established to ensure the broadness of the deep domain velocity model set covering the geological structure characteristics of the actual processing work area.

[0231] In the embodiment, the deep domain velocity model set is defined according to formula (1):

[0232] V depth = Γ(V m , N dip , M laye r, D abnormal )

[0233] V m ∈[1000, 7000], N dip ∈[-35°, 35°] and different with layers,

[0234] M layer ∈[1, 14], D abnormal ∈[0, 1]

[0235] wherein, V depth represents the total velocity parameter of the current deep domain velocity model, the function Γ is an abstracted velocity model representation, V m represents the layer velocity of the Nth layer of the current deep domain velocity model, wherein N is between 0-M layer , Mlayer is the number of layers, the layer dip angle is N dip , wherein N dip is between -35 and 35 degrees, and D abnormalIndicates whether there is an abnormal body in the layer velocity. That is, according to the number of layers Mlayer, the dip angle Ndip, the velocity parameter Vm, the depth parameter Dabnormal of the generated velocity abnormal body, and the generated depth domain velocity model. Based on the dip angle, the velocity range, the geological abnormal body, and the number of layers, the depth domain velocity model is established, so that the depth domain velocity model set is more comprehensive, and adapts to the geological complex situation of the domestic seismic exploration work area.

[0236] In one embodiment, the computer program, when executed by the processor, further implements the following steps:

[0237] Step 111, two-way wave equation forward modeling is performed on the depth domain velocity model set to obtain synthetic seismic data.

[0238] Step 112, gain compensation processing is performed on the synthetic seismic data to obtain gather data.

[0239] In this embodiment, the two-way wave equation is an acoustic two-way wave equation, and the depth domain velocity model set is processed by the acoustic two-way wave equation to prepare synthetic seismic data through forward modeling strategy.

[0240] In this embodiment, the step of performing gain compensation processing on the synthetic seismic data to obtain gather data includes: after gain compensation processing is performed on the synthetic seismic data, compensated seismic data is obtained; and the compensated seismic data is sorted in a CMP arrangement format to obtain gather data. It should be understood that when the synthetic seismic data is prepared to obtain stacking velocity spectrum data, the input data is sorted in a CMP arrangement format.

[0241] In one embodiment, the computer program, when executed by the processor, further implements the following steps:

[0242] Step 121, based on the stacking velocity formula, features are extracted from the gather data to obtain gather features, and the gather features are converted into stacking velocity spectrum data;

[0243] Step 122, the stacking velocity spectrum data is segmented and normalized to obtain sample set data.

[0244] In this embodiment, the gather data is feature-extracted according to formula (2):

[0245]

[0246] wherein Spec amp represents the prepared stacking velocity spectrum data, v represents the current scanning velocity, and the range of v is v start v endbetween the two, the specific value of v is determined according to the velocity range of the depth domain velocity model, dt is the data sampling point interval, ntrace represents the total number of seismic data, wherein CMP represents the current gather data to be processed. nt is the sampling point number, it is the index of the sampling point, ntrace is the total number of gather data, itrace is the index of the trace, v staxt is the start of the scanning speed, v end is the end of the scanning speed, the velocity parameter dt is the sampling interval, v is the current scanning speed, X offset is the offset length corresponding to the current seismic trace. In this embodiment, the velocity range of the depth domain velocity model is 1000m / s-7350m / s. In this embodiment, the data sampling point interval dt is 0.5ms.

[0247] In this embodiment, the stacking velocity spectrum data is segmented and normalized according to formula (3) to obtain sample set data;

[0248] Spec norm (v, it) = Spec amp (v, it) / Max(SpeC amp (v, it))it∈[t_start, t_end] (3)

[0249] wherein Spec norm represents the sample set data, t_start and t_end depend on the time window size. In this embodiment, the time window length is 10 sampling points.

[0250] In one embodiment, the computer program is executed by the processor to further implement the following steps:

[0251] Step 131, extracting the gather point velocity from the depth domain velocity model set to obtain the interval velocity curve;

[0252] Step 132, converting the interval velocity curve into the time domain root mean square velocity curve;

[0253] Step 133, normalizing the time domain root mean square velocity curve to obtain the label set data.

[0254] In this embodiment, the CMP center point interval velocity curve is extracted from the depth domain velocity model.

[0255] In this embodiment, based on the DIX formula, the interval velocity curve is converted into the time domain root mean square velocity curve.

[0256] In one embodiment, the step of normalizing the time domain root mean square velocity curve comprises:

[0257] The time-domain root-mean-square velocity curve is normalized based on maximum and minimum velocities.

[0258] In this embodiment, the time-domain root-mean-square velocity curve is normalized based on formula (4):

[0259] V = (V-V min ) / (V max -V min )

[0260] wherein V represents a seismic velocity model, V min and V max represent a set normalized velocity range, V min represents a normalized minimum velocity, and V max represents a normalized maximum velocity. In this embodiment, the normalized velocity range is 1000 m / s-7350 m / s, V min is 1000 m / s, and V max is 7350 m / s.

[0261] In this embodiment, the maximum and minimum velocities in the maximum and minimum velocity normalization are determined according to the characteristics of the work area.

[0262] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by instructing related hardware based on a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above-mentioned embodiments. Any reference to memory, storage, database or other medium used in the embodiments provided by the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0263] Any combination of the technical features in the above embodiments can be made, and for the sake of brevity, not all possible combinations are described above, however, as long as the combination of the technical features does not exist in contradiction, it shall be considered within the scope of the present disclosure.

[0264] The above embodiments only express several implementation manners of the present application, and the description is relatively specific and detailed, but it shall not be understood as a limitation on the patent scope of the present application. It shall be pointed out that, for ordinary skilled persons in the art, several modifications and improvements can be made without departing from the concept of the present application, and these shall be within the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the appended claims.

Claims

1. A method of seismic survey velocity analysis, characterized by, The method comprises the following steps: processing a depth domain velocity model set based on a two-way wave equation to obtain gather data; processing the gather data based on a stacking velocity spectrum production formula to obtain sample set data; extracting a time domain root mean square velocity curve from the depth domain velocity model set to obtain label set data; training a preset deep learning model based on the sample set data and the label set data to obtain a seismic exploration velocity modeling system for seismic exploration velocity analysis.

2. The method of claim 1, wherein, The method further comprises the following steps before the step of processing the depth domain velocity model set based on the two-way wave equation to obtain the gather data: obtaining geological conditions of multiple seismic exploration work areas, analyzing the geological conditions of each of the seismic exploration work areas to obtain geological structure features, and establishing the depth domain velocity model set based on the geological structure features.

3. The method of claim 1, wherein, The method further comprises the following steps in the step of processing the depth domain velocity model set based on the two-way wave equation to obtain the gather data: performing two-way wave equation forward modeling on the depth domain velocity model set to obtain synthetic seismic data; performing gain compensation processing on the synthetic seismic data to obtain the gather data.

4. The method of claim 1, wherein, The method further comprises the following steps in the step of processing the gather data based on the stacking velocity spectrum production formula to obtain the sample set data: extracting features from the gather data based on the stacking velocity spectrum production formula to obtain gather features, and converting the gather features into stacking velocity spectrum data; performing segmented normalization processing on the stacking velocity spectrum data to obtain the sample set data.

5. The method according to any one of claims 1 to 4, characterized in that, The method further comprises the following steps in the step of extracting the time domain root mean square velocity curve from the depth domain velocity model set to obtain the label set data: extracting gather point velocities from the depth domain velocity model set to obtain a layer velocity curve; converting the layer velocity curve into the time domain root mean square velocity curve; performing normalization processing on the time domain root mean square velocity curve to obtain the label set data.

6. The method of claim 5, wherein, The method further comprises the following step in the step of performing normalization processing on the time domain root mean square velocity curve: performing normalization processing on the time domain root mean square velocity curve based on maximum and minimum velocities.

7. A seismic survey velocity analysis device, characterized by, The method comprises the following steps: a gather data extraction module that processes a depth domain velocity model set based on a two-way wave equation to obtain gather data; a sample set data acquisition module that processes the gather data based on a stacking velocity spectrum production formula to obtain sample set data; a label set data acquisition module that extracts a time domain root mean square velocity curve from the depth domain velocity model set to obtain label set data; a model training module that trains a preset deep learning model based on the sample set data and the label set data to obtain a seismic exploration velocity modeling system for seismic exploration velocity analysis.

8. The apparatus of claim 7, wherein, The seismic exploration velocity analysis device comprises: The speed model establishing module is configured to acquire the geological conditions of multiple seismic exploration work areas, analyze the geological conditions of each of the seismic exploration work areas to obtain geological structure features, and establish the depth domain speed model set based on the geological structure features. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-8 when the computer program is executed by the processor. The computer program, when executed by the processor, implements the steps of the method of any one of claims 1 to 6.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program, when executed by the processor, implements the steps of the method of any one of claims 1 to 6.

Images