Construction of first-arrival picking model and method and apparatus for first-arrival picking

Abstract

This disclosure relates to a method and apparatus for constructing a first arrival picking model and for first arrival picking. The method includes: acquiring multiple seismic sampling data points corresponding to multiple single-shot excitations for each target sampling point location in a low signal-to-noise ratio area of ​​the target exploration area; vertically stacking the multiple seismic sampling data points to obtain stacked seismic data, and performing first arrival picking on the stacked seismic data to obtain the target first arrival; using the seismic sampling data corresponding to the multiple target sampling point locations as training input to an artificial intelligence model, the artificial intelligence model outputting the first arrival picking results, and using the target first arrival as training labels to supervise the training of the artificial intelligence model, the trained artificial intelligence model serving as the first arrival picking model. The first arrival picking model has the advantage of strong applicability to different types of noise and different geological conditions in different work areas, and can improve the accuracy of first arrival picking.

Description

Technical Field

[0001] This disclosure relates to the fields of seismic data processing and artificial intelligence technology, and in particular to a method and apparatus for constructing an initial arrival picking model and initial arrival picking. Background Technology

[0002] In the field of oil and gas exploration, the main processes involved are seismic data acquisition, processing, and interpretation. Seismic data processing primarily assists in the interpretation of seismic data (also known as seismic data itself), and includes static correction, denoising, deconvolution, dynamic-static correction, velocity analysis, stacking, and migration. During seismic data processing, it is essential to first eliminate the distortion caused by near-surface low-velocity zones on the travel time of reflected waves. The accuracy of first-arrival picking is crucial for establishing surface structure models through first-arrival inversion, which in turn affects the accuracy of shallow velocity model construction.

[0003] In the process of realizing the present invention, the inventors discovered at least the following technical problems in the related technologies: In the related technologies, as the accuracy of exploration instruments improves, the amount of seismic single-shot data increases exponentially, and the workload of first arrival picking also increases. Some solutions have begun to consider using artificial intelligence models for training to achieve first arrival picking. However, most current artificial intelligence models select some data from a single shot as samples for training. The first arrival picking model obtained in this way has poor accuracy in first arrival picking for seismic data corresponding to some geological areas with low signal-to-noise ratio. Summary of the Invention

[0004] To solve the above-mentioned technical problems, or at least partially solve them, embodiments of this disclosure provide a method and apparatus for constructing an initial arrival picking model and initial arrival picking.

[0005] In a first aspect, embodiments of this disclosure provide a method for constructing a first arrival picking model. The method includes: acquiring multiple seismic sampling data points corresponding to multiple single-shot excitations for each target sampling point location in a low signal-to-noise ratio area of ​​the target exploration area; vertically stacking the multiple seismic sampling data points to obtain stacked seismic data, and performing first arrival picking on the stacked seismic data to obtain the target first arrival; using the seismic sampling data corresponding to the multiple target sampling point locations as training input to an artificial intelligence model, the artificial intelligence model outputting the first arrival picking result; using the target first arrival as training labels to supervise the training of the artificial intelligence model, and the trained artificial intelligence model serving as the first arrival picking model.

[0006] In some embodiments, the aforementioned multiple seismic sampling data are acquired by: determining candidate sampling points corresponding to different noise types within the aforementioned first-arrival low signal-to-noise ratio region based on existing seismic data of the target exploration area; determining the target sampling point location and corresponding number of acquisitions for point-to-multiple sampling based on the location of the candidate sampling points, existing first-arrival signal-to-noise ratio information, and preset coverage information of noise types; and acquiring multiple single-shot excitation results at the target sampling point location within the aforementioned low signal-to-noise ratio region according to the aforementioned number of acquisitions, thereby obtaining multiple seismic sampling data.

[0007] In some embodiments, the aforementioned preset coverage information is used to indicate a preset number or preset proportion of noise types to be covered by the training samples. Determining the target sampling point location and corresponding sampling count for one-point-multiple-sampling based on the location of the candidate sampling points, the existing first-to-noise ratio (SNR) information, and the preset coverage information for the noise types includes: determining the location of all or part of the candidate sampling points corresponding to the noise types as the target sampling point location for one-point-multiple-sampling based on the preset coverage information; determining the minimum number of stacking operations required to satisfy the condition that the SNR exceeds a threshold value after vertical stacking based on the existing SNR information of the target sampling point location; the threshold value being the division threshold corresponding to the low-to-high SNR region and the high-to-low SNR region; randomly generating a sampling count that satisfies the minimum number of stacking operations, or using the number of input operations exceeding the minimum number of stacking operations as the sampling count.

[0008] In some embodiments, the number of times each target sampling point is superimposed satisfies the following expression:

[0009]

[0010] Among them, S p / N p S / N represents the initial-to-noise ratio after vertical stacking; S / N represents the existing initial-to-noise ratio corresponding to the current target sampling point position; P represents the number of stacking times; C represents the boundary value corresponding to the division between the initial-to-low signal-to-noise ratio region and the initial-to-high signal-to-noise ratio region.

[0011] By solving the above expression, the minimum value of P is obtained as the minimum number of superpositions.

[0012] In some embodiments, the method further includes: dividing the target exploration area into a low initial arrival signal-to-noise ratio (SNR) region and a high initial arrival signal-to-noise ratio (SNR) region. The low initial arrival SNR region and the high initial arrival SNR region are divided as follows: based on existing seismic data of the target exploration area, the total initial arrival energy and the total noise energy are determined; the ratio of the total initial arrival energy to the total noise energy is calculated to obtain the overall SNR; a boundary value is determined based on the overall SNR; the boundary value is the threshold value corresponding to the division of the low initial arrival SNR region and the high initial arrival SNR region; and the regions are divided according to the boundary value and the regional SNR of each region to obtain the low initial arrival SNR region and the high initial arrival SNR region.

[0013] Secondly, embodiments of this disclosure provide a method for first-arrival picking. The method includes: acquiring seismic data to be processed; inputting the seismic data to be processed into a pre-constructed first-arrival picking model, and outputting the corresponding first-arrival picking result; wherein the first-arrival picking model is constructed using the method described above for constructing a first-arrival picking model.

[0014] Thirdly, embodiments of this disclosure provide an apparatus for constructing a first-arrival picking model. The apparatus includes a data acquisition module, a training label calculation module, and a training module. The data acquisition module is used to acquire multiple seismic sampling data obtained from multiple single-shot excitations corresponding to each target sampling point location in a low signal-to-noise ratio area of ​​the target exploration area. The training label calculation module is used to vertically stack the multiple seismic sampling data to obtain stacked seismic data, and to perform first-arrival picking on the stacked seismic data to obtain the target first arrival. The training module is used to input the seismic sampling data corresponding to the multiple target sampling point locations as training input to an artificial intelligence model. The artificial intelligence model outputs the first-arrival picking result, and uses the target first arrival as training labels to supervise the training of the artificial intelligence model. The trained artificial intelligence model serves as the first-arrival picking model.

[0015] Fourthly, embodiments of this disclosure provide an apparatus for first arrival picking. The apparatus includes a data acquisition module and a data processing module. The data acquisition module acquires seismic data to be processed. The data processing module inputs the seismic data to be processed into a pre-constructed first arrival picking model and outputs a corresponding first arrival picking result; wherein the first arrival picking model is constructed using a method for constructing first arrival picking models or using the apparatus for constructing first arrival picking models.

[0016] Fifthly, embodiments of this disclosure provide an electronic device. The electronic device includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, communication interface, and memory communicate with each other via the communication bus; the memory stores computer programs; and the processor, when executing the program stored in the memory, implements the method for constructing an initial arrival picking model or the initial arrival picking method as described above.

[0017] Sixthly, embodiments of this disclosure provide a computer-readable storage medium. The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method for constructing an initial arrival picking model or the initial arrival picking method as described above.

[0018] The technical solutions provided in the embodiments of this disclosure have at least some or all of the following advantages:

[0019] In the low signal-to-noise ratio area of ​​the target exploration region, multiple seismic sampling data are obtained by single-shot excitation at each target sampling point location, achieving multiple sampling from a single point. Since these multiple seismic sampling data carry noise from the real geological environment and can reflect the impact of various noise types on seismic data, the seismic sampling data corresponding to multiple target sampling points serves as input to the artificial intelligence model, enabling the samples to reflect the geological structure and signal interference types corresponding to the target exploration region. Simultaneously, the target first arrival (which can be the first arrival wave or first arrival time) corresponding to the same target sampling point location is obtained through multiple seismic sampling data collected at that location. Vertical stacking and first-arrival picking result in stacked seismic data that enhances the signal while reducing the relative proportion of noise, thus improving the signal-to-noise ratio (SNR) of the stacked seismic data. When seismic sampling data corresponding to the same target sampling point is used as training input, the corresponding training label is the target first arrival at that sampling point. This is more accurate than training labels obtained by extracting the first arrival of each individual seismic sampling data point, thereby improving the effectiveness of supervised training. This allows the trained first-arrival picking model to be applicable and relatively accurately pick up the first arrival wave or first arrival time in real-world low SNR regions and for various noise types. Therefore, the aforementioned method for constructing or obtaining a first-arrival picking model has the advantage of strong applicability to different types of noise and geological conditions in different work areas, and can improve the accuracy of first-arrival picking. Attached Figure Description

[0020] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0021] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, those skilled in the art can obtain other drawings based on these drawings without creative effort.

[0022] Figure 1 A flowchart illustrating a method for constructing an initial arrival picking model according to an embodiment of the present disclosure is shown schematically;

[0023] Figure 2 A schematic diagram illustrating the acquisition process of multiple seismic sampling data according to an embodiment of the present disclosure is shown.

[0024] Figure 3 This diagram illustrates a method for vertically stacking multiple seismic sampling data according to an embodiment of the present disclosure to obtain stacked seismic data.

[0025] Figure 4 The diagram schematically illustrates (a) the original signal-to-noise ratio of seismic sampling data corresponding to a single shot excitation according to an embodiment of the present disclosure, and (b) a comparison diagram of the signal-to-noise ratio of the superimposed seismic data obtained after 12 superpositions.

[0026] Figure 5 A schematic diagram illustrating the acquisition of a target first arrival from stacked seismic data according to an embodiment of the present disclosure is shown.

[0027] Figure 6 The illustration shows a schematic diagram of (a) the target first arrival obtained by first arrival picking of superimposed seismic data, and (b) the first arrival of the target corresponding to the same target sampling point position as training labels for first arrival picking supervised training of single-shot-excited seismic sampling data according to an embodiment of the present disclosure.

[0028] Figure 7 A schematic block diagram of an electronic device provided in an embodiment of the present disclosure is shown. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this disclosure. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.

[0030] During the research and development process, at least the following technical problems were found in the relevant technologies: In the relevant technologies, as the accuracy of exploration instruments improves, the amount of seismic single-shot data increases exponentially, and the workload of first arrival picking also increases. Some solutions have begun to consider using artificial intelligence models for training to achieve first arrival picking. However, most current artificial intelligence models select some data from a single shot as samples for training. The first arrival picking model obtained in this way has poor accuracy for seismic data corresponding to some geological areas with low signal-to-noise ratio.

[0031] In view of this, in order to improve the accuracy and applicability of the constructed first-arrival picking model in picking first arrivals, and to provide control and verification data for the first-arrival tomography inversion model required for static correction and pre-stack depth migration imaging of onshore oil seismic exploration data processing, this disclosure provides a method and apparatus for constructing a first-arrival picking model and picking first arrivals. This mainly involves achieving multiple sampling at a single point in the low signal-to-noise ratio area of ​​the target exploration area, vertically stacking multiple seismic sampling data collected at the same target sampling point location, and picking the target first arrival. This target first arrival is then used as multiple input data for the same target sampling point location. All have corresponding training labels; since the seismic sampling data corresponding to multiple target sampling points are used as input to the artificial intelligence model, the samples can reflect the geological structure and signal interference type of the target exploration area. At the same time, the training labels are more accurate results obtained by first arrival extraction from superimposed seismic data with high signal-to-noise ratio. Compared with the training labels obtained by first arrival extraction from each seismic sampling data separately, the training labels are more accurate, thereby improving the effect of supervised training. This allows the trained first arrival picking model to be applicable and relatively accurate in picking first arrival waves or first arrival times in real low signal-to-noise ratio areas and for various noise types.

[0032] The following detailed description is provided in conjunction with specific examples.

[0033] This disclosure provides a first exemplary embodiment of a method for constructing an initial arrival picking model. The method of this embodiment can be executed by an electronic device with computing capabilities.

[0034] Figure 1 A flowchart illustrating a method for constructing an initial arrival picking model according to an embodiment of the present disclosure is shown schematically.

[0035] Reference Figure 1 As shown, the method for constructing an initial arrival picking model provided in this embodiment includes the following steps: S110, S120 and S130.

[0036] In step S110, for the initial low signal-to-noise ratio area of ​​the target exploration area, multiple seismic sampling data corresponding to the location of each target sampling point are obtained from multiple single-shot excitations.

[0037] The target exploration area is the object to be explored. Some seismic data acquisition and processing work can be carried out in advance for the target exploration area. Based on the existing seismic data of the target exploration area, the target exploration area can be divided into the initial low signal-to-noise ratio area and the initial high signal-to-noise ratio area.

[0038] In some implementation scenarios, in step S110, the aforementioned electronic device and the seismic mining equipment are independent devices, capable of data transmission between them. The electronic device acquires multiple seismic sampling data points corresponding to each target sampling point location obtained from multiple single-shot firings from the seismic mining equipment. In other implementation scenarios, in step S110, the aforementioned electronic device integrates a seismic data acquisition module, enabling it to acquire multiple seismic sampling data points corresponding to each target sampling point location obtained from multiple single-shot firings during the seismic firing process.

[0039] In some embodiments, prior to step S110, the method further includes the following step: dividing the target exploration area into a low signal-to-noise ratio region and a high signal-to-noise ratio region.

[0040] The low signal-to-noise ratio (SNR) region refers to the area in the target exploration region where the ratio of initial arrival energy to noise energy is less than the threshold value; the high SNR region refers to the area in the target exploration region where the ratio of initial arrival energy to noise energy exceeds the threshold value. These threshold values ​​are the dividing lines for the low and high SNR regions.

[0041] The aforementioned regions with low signal-to-noise ratios and high signal-to-noise ratios are divided as follows:

[0042] Based on existing seismic data of the target exploration area, the total energy E of the first arrival wave was determined. S Total energy of noise E n ;

[0043] Calculate the total energy E of the first arrival wave mentioned above. S And the total noise energy E mentioned above n The ratio of the two values ​​is used to obtain the overall signal-to-noise ratio.

[0044] Based on the overall signal-to-noise ratio, the boundary value is determined; the boundary value is the division threshold corresponding to the initial low signal-to-noise ratio region and the initial high signal-to-noise ratio region.

[0045] Based on the above-mentioned boundary values ​​and the regional signal-to-noise ratio of each region, the regions are divided into regions from the initial low signal-to-noise ratio region and the initial high signal-to-noise ratio region.

[0046] In some embodiments, the above-mentioned boundary value is related to the overall signal-to-noise ratio of the target exploration area (as an average result). For example, the larger the overall signal-to-noise ratio of the target exploration area, the larger the corresponding boundary value; the smaller the overall signal-to-noise ratio of the target exploration area, the smaller the corresponding boundary value.

[0047] Figure 2 A schematic diagram illustrating the acquisition process of multiple seismic sampling data according to an embodiment of the present disclosure is shown.

[0048] In some embodiments, refer to Figure 2 As shown, the above-mentioned multiple seismic sampling data were obtained through the following methods:

[0049] In step S210, based on the existing seismic data of the target exploration area, candidate sampling points corresponding to different noise types in the aforementioned initial low signal-to-noise ratio area are determined.

[0050] For example, by using existing seismic data of the target exploration area, we can identify different noise types based on the differences in noise distribution patterns, and thus determine the candidate sampling points corresponding to different noise types.

[0051] In step S220, based on the location of the candidate sampling points, the existing first-arrival signal-to-noise ratio information, and the preset coverage information of the noise type, the location of the target sampling point and the corresponding number of samplings for performing one-point multi-sampling are determined.

[0052] In some embodiments, the aforementioned preset coverage information is used to indicate a preset quantity or preset proportion of noise types to be covered by the training samples. In some scenarios, the aforementioned preset quantity or preset proportion can be set according to the accuracy requirements for identifying initial arrivals in the initial arrival picking model to be built.

[0053] In step S220 above, based on the location of the candidate sampling points, the existing first-arrival signal-to-noise ratio information, and the preset coverage information of the noise type, the location of the target sampling point for performing one-point multi-sampling and the corresponding number of samplings are determined, including:

[0054] Based on the aforementioned preset coverage information, the locations of candidate sampling points corresponding to all or part of the noise types are determined as the target sampling point locations for performing one-point multi-sampling.

[0055] Based on the existing first-to-first signal-to-noise ratio (SNR) information of the target sampling point location, determine the minimum number of stacking operations required to satisfy the condition that the first-to-first SNR exceeds the threshold value after vertical stacking; the threshold value is the division threshold corresponding to the first-to-first low SNR region and the first-to-first high SNR region; for example, in some embodiments, the threshold value is 1.

[0056] Randomly generate the number of collections that meets the minimum number of stackings mentioned above, or use the number of inputs that exceeds the minimum number of stackings mentioned above as the number of collections.

[0057] In this embodiment, the number of acquisitions can be obtained by random generation or by user input, and this number of acquisitions is greater than or equal to the minimum number of stackings. This effectively ensures that the signal-to-noise ratio of the stacked seismic data obtained after vertical stacking of the seismic data corresponding to multiple acquisitions at one point (i.e., stacking multiple single-shot data corresponding to the same location point) is improved. As a result, the target first arrival waveform or target first arrival time obtained by first arrival picking based on the stacked seismic data is clear and accurate.

[0058] In some embodiments, the number of times each target sampling point is superimposed satisfies the following expression:

[0059]

[0060] Among them, S p / N p S represents the initial signal-to-noise ratio after vertical stacking; S / N represents the existing initial signal-to-noise ratio corresponding to the current target sampling point position; P represents the number of stackings; C represents the boundary value corresponding to the division of the initial low signal-to-noise ratio region and the initial high signal-to-noise ratio region (for example, it can be taken as 1); by solving the above expression, the minimum value of P is obtained as the minimum number of stackings, and values ​​greater than or equal to the minimum number of stackings can be used as the value of the number of samplings J.

[0061] In some embodiments, the first-arrival signal-to-noise ratio is calculated using the following formula:

[0062]

[0063] Among them, E S E represents the total first-arrival energy obtained from single-shot seismic data (e.g., the first shot data) that is free from noise interference. n The total noise energy corresponds to the first arrival of single-shot seismic data affected by noise (e.g., the second shot data); M is the number of channels within the time window, T is the time window size, Δt is the sampling duration, i is the number of samples within the time window, and n is the total noise energy. i Let s be the noise energy of the i-th sample point. i Let be the initial arrival energy of the i-th sample point.

[0064] In step S230, at the target sampling point location in the aforementioned low signal-to-noise ratio region, multiple single-shot excitation results are collected according to the aforementioned number of collections to obtain multiple seismic sampling data.

[0065] By performing multiple excitations at the target sampling point location, multiple corresponding seismic sampling data can be obtained at the same target sampling point location. For the entire low signal-to-noise ratio region, there are multiple target sampling point locations; for example, the sequence corresponding to the target sampling point locations can be represented as {L1, L2, ..., L...}.k , ..., L K}, L k Let L represent the location of the k-th target sampling point; k represents the index of the target sampling point location, k = 1 to K, and K represents the total number of target sampling point locations, K ≥ 2 and is an integer; thus, multiple sets of seismic sampling data can be obtained, where each target sampling point location L k Corresponding to a set of seismic sampling data {s k1 s k2 …s kr …s kRk}, s kr L represents the position of the kth target sampling point. k The corresponding r-th seismic sampling data; Rk represents the location L of the target sampling point. k The total number of corresponding seismic sampling data is J, which is the number of sampling times for the corresponding target sampling point location, where Rk ≥ 2 and is an integer. For different target sampling point locations L... k The total number of corresponding seismic sampling data Rk can be the same or different, depending on the number of samplings determined in the implementation details of the aforementioned step S220.

[0066] In the initial low signal-to-noise ratio area of ​​the target exploration area, multiple single-shot excitations are collected for each target sampling point to obtain multiple seismic sampling data, realizing multiple sampling at one point. Since the above multiple seismic sampling data carry noise from the real geological environment and can reflect the impact of various types of noise on seismic data, the seismic sampling data corresponding to multiple target sampling points can be used as input to the artificial intelligence model, enabling the samples to reflect the geological structure and signal interference type of the target exploration area.

[0067] In step S120, the multiple seismic sampling data are vertically superimposed to obtain superimposed seismic data, and the first arrival of the target is picked up based on the superimposed seismic data.

[0068] In seismic exploration, the moment when a seismic wave front arrives at an observation point and the detector at that point detects the vibration of a particle is called the first arrival time of the wave, or simply the first arrival. The first wave to arrive in the seismic record is called the first arrival wave, and subsequent waves appear against the background of vibration and are called subsequent arrival waves.

[0069] Figure 3 This diagram schematically illustrates the vertical stacking of multiple seismic sampling data according to an embodiment of the present disclosure to obtain stacked seismic data. To illustrate the overall stacking effect, Figure 3Details of each plot (such as the values ​​of the coordinate axes) are not shown. The horizontal and vertical coordinates of each seismic sampling data before stacking and the stacked seismic data obtained by vertical stacking are the same (i.e., the results correspond to the same coordinate system). According to common knowledge in this field, the horizontal axis represents the number of traces and the vertical axis represents time. Figure 4 The diagram schematically illustrates (a) the original signal-to-noise ratio of seismic sampling data corresponding to a single shot excitation according to an embodiment of the present disclosure, and (b) a comparison diagram of the signal-to-noise ratio of the superimposed seismic data obtained after 12 superpositions.

[0070] Reference Figure 3 , Figure 4 As shown in (a) and (b), in this embodiment, the number of samplings J is set to be equal to the minimum number of superpositions P. min Multiple seismic sampling data corresponding to the same target sampling point are described as data corresponding to repeated acquisition shot 1 to repeated acquisition shot P. The signal-to-noise ratio of the original single shot before stacking is 0.523, and the signal-to-noise ratio of the stacked seismic data after stacking P times is greater than or equal to 1. For example, refer to Figure 4 As shown in (b), the signal-to-noise ratio of the stacked seismic data obtained by overlaying 12 single-shot data is 1.657.

[0071] Figure 5 A schematic diagram illustrating the acquisition of a target first arrival from stacked seismic data according to an embodiment of the present disclosure is shown.

[0072] Reference Figure 5 As shown, the superimposed seismic data improves upon the signal-to-noise ratio, from... Figure 5 As can be seen from the data, the boundary of the corresponding first arrival wave is relatively clear, and the first arrival can be distinguished. Therefore, the target first arrival extracted from the stacked seismic data is more accurate than the first arrival extracted from each seismic sampling data corresponding to the same target sampling point location.

[0073] By vertically stacking multiple seismic sampling data, for example, for the location L1 of the first target sampling point, the corresponding seismic sampling data s are stacked... 11 s 12 …s 1r …s 1R1 Vertical stacking is performed to obtain stacked seismic data s1 stack Therefore, regarding the superimposed seismic data s1 stack Perform initial arrival picking to obtain the initial arrival X0 of the target at the first target sampling point position L1. 1 The specific initial arrival of the target can be either the initial arrival wave of the target or the initial arrival time of the target.

[0074] By vertically stacking multiple seismic sampling data at the target sampling point location, the signal-to-noise ratio of a single shot is improved, reaching the original level. (The square root of J represents the number of times data was collected.)

[0075] Similarly, for the k-th target sampling point position L k The corresponding seismic sampling data s k1 s k2 …s kr …s kRk Vertical stacking is performed to obtain stacked seismic data s k stack. ; Regarding the superimposed seismic data s k stack. Perform initial pickup to obtain the position L of the k-th target sampling point. k The initial target is X0 k .

[0076] In step S130, the seismic sampling data corresponding to the locations of multiple target sampling points are used as training inputs and input into the artificial intelligence model. The artificial intelligence model outputs the first arrival picking results. The first arrival of the targets is used as training labels to supervise the training of the artificial intelligence model. The trained artificial intelligence model is used as the first arrival picking model.

[0077] Figure 6 The illustration schematically shows a process according to an embodiment of the present disclosure, in which (a) the target first arrival is obtained by first arrival picking from stacked seismic data; and (b) the target first arrival corresponding to the same target sampling point location is used as a training label to perform supervised training on first arrival picking of seismic sampling data excited by a single shot.

[0078] Combination Figure 5 , Figure 6 As shown in (a) and (b), the target first arrival (which can be the first arrival wave or the first arrival time) corresponding to the same target sampling point is obtained by vertically superimposing multiple seismic sampling data collected at that location and picking the first arrival. The superimposed seismic data obtained after vertical superposition will enhance the signal and reduce the relative proportion of noise, thus improving the signal-to-noise ratio of the superimposed seismic data. In this way, when the seismic sampling data corresponding to the same target sampling point is used as training input, the corresponding training label is the target first arrival corresponding to that target sampling point. This is more accurate than the training label obtained by extracting the first arrival of each seismic sampling data separately, thereby improving the effect of supervised training.

[0079] When training the artificial intelligence model, the training input sample set consists of seismic sampling data corresponding to all target sampling point locations. This seismic sampling data includes samples with one or more noise types, as well as samples without noise. Specifically, in each training process, the input is a single seismic sampling data point corresponding to a specific target sampling point location (e.g., the location L of the k-th target sampling point). k Corresponding seismic sampling data s k1 s k2 …s kr …s kRk One of them), the corresponding output obtains the corresponding initial arrival picking result. For multiple inputs corresponding to the same target sampling point position, the same target initial arrival X0 is used. k Supervised training is conducted; during the training iteration process, the parameters of the artificial intelligence model are adjusted until the number of training iterations reaches a set value or the corresponding loss function is lower than a set threshold, at which point the training is considered complete.

[0080] The difference between the initial arrival picking result obtained from the output and the corresponding training label (target initial arrival) is used as the loss function.

[0081] In embodiments including steps S110 to S130 above, in the low signal-to-noise ratio area of ​​the first arrival in the target exploration area, multiple seismic sampling data are obtained by multiple single-shot excitations for each target sampling point location, achieving multiple sampling from one point. Since the above multiple seismic sampling data carry noise from the real geological environment and can reflect the impact of various real noise types on seismic data, the seismic sampling data corresponding to multiple target sampling point locations serves as input to the artificial intelligence model, enabling the samples to reflect the geological structure and signal interference type corresponding to the target exploration area. Simultaneously, since the target first arrival (which can be the first arrival wave or first arrival time) corresponding to the same target sampling point location is obtained by targeting that location... Multiple seismic sampling data are vertically superimposed and first-arrival picking is performed. This vertical superposition enhances the signal while reducing the relative proportion of noise, thus improving the signal-to-noise ratio (SNR) of the superimposed seismic data. When seismic sampling data corresponding to the same target sampling point is used as training input, the corresponding training label is the target first-arrival at that point. This is more accurate than training labels obtained by extracting the first-arrival of each individual seismic sampling data point, thereby improving the effectiveness of supervised training. The trained first-arrival picking model is applicable and relatively accurate in picking first-arrival waves or first-arrival times in real-world low SNR regions and for various noise types. Therefore, the aforementioned method for constructing or obtaining a first-arrival picking model is highly applicable and improves the accuracy of first-arrival picking. For example, when the trained first-arrival picking model is applied to other work areas, the pre-trained sample set contains various noise types, effectively improving the applicability of the first-arrival picking model to different noise types and geological conditions in different work areas.

[0082] A second exemplary embodiment of this disclosure provides a method for initial arrival picking.

[0083] The above method includes: acquiring seismic data to be processed; inputting the seismic data to be processed into a pre-constructed first arrival picking model, and outputting the corresponding first arrival picking result; wherein the first arrival picking model is constructed using the above method for constructing the first arrival picking model.

[0084] In this embodiment, the seismic data to be processed can be the same as or different from the work area where the target exploration area is located during the training phase. That is, the work area corresponding to the seismic data to be processed can be the data corresponding to a work area that has not yet been explored.

[0085] Since the pre-trained sample set contains various types of noise, it can effectively improve the applicability of the first arrival picking model to different types of noise and different geological conditions in different work areas. Therefore, the above-mentioned first arrival picking model also has good applicability and meets the expected first arrival picking accuracy when picking the first arrival of seismic records in unexplored work areas.

[0086] A third exemplary embodiment of this disclosure provides an apparatus for constructing an initial arrival picking model.

[0087] In this embodiment, the apparatus for constructing the initial picking model includes: a data acquisition module, a training label calculation module, and a training module.

[0088] The aforementioned data acquisition module is used to acquire multiple seismic sampling data obtained from multiple single-shot excitations for each target sampling point location in the initial low signal-to-noise ratio area of ​​the target exploration area.

[0089] The training label calculation module is used to vertically stack the multiple seismic sampling data to obtain stacked seismic data, and to pick the first arrival of the target from the stacked seismic data.

[0090] The training module described above is used to take the seismic sampling data corresponding to the locations of multiple target sampling points as training inputs to the artificial intelligence model. The artificial intelligence model outputs the first arrival picking results, and the first arrival of the targets is used as training labels to supervise the training of the artificial intelligence model. The trained artificial intelligence model is used as the first arrival picking model.

[0091] More details and beneficial effects included in this embodiment can be found in the relevant description of the first embodiment, which will not be repeated here.

[0092] A fourth exemplary embodiment of this disclosure provides an apparatus for initial arrival pickup.

[0093] The initial pickup device in this embodiment includes a data acquisition module and a data processing module.

[0094] The data acquisition module described above is used to acquire earthquake data to be processed.

[0095] The aforementioned data processing module is used to input the aforementioned seismic data to be processed into a pre-constructed first arrival picking model and output the corresponding first arrival picking result; wherein, the aforementioned first arrival picking model is constructed by the method for constructing a first arrival picking model or by the aforementioned device for constructing a first arrival picking model.

[0096] More details and beneficial effects contained in this embodiment can be found in the relevant descriptions of the first and second embodiments, which will not be repeated here.

[0097] Any number of functional modules included in the apparatus provided in the third and fourth embodiments above can be combined into one module, or any one of these modules can be split into multiple modules. Alternatively, at least part of the functionality of one or more of these modules can be combined with at least part of the functionality of other modules and implemented in one module. At least one of the functional modules included in the apparatus provided in the third and fourth embodiments above can be at least partially implemented as hardware circuits, such as field-programmable gate arrays (FPGAs), programmable logic arrays (PLAs), systems-on-a-chip, systems-on-a-substrate, systems-on-package, application-specific integrated circuits (ASICs), or implemented by any other reasonable means of integrating or packaging circuits, or implemented by any one of software, hardware, and firmware, or by any appropriate combination of any of these three implementation methods. Alternatively, at least one of the functional modules included in the apparatus provided in the third and fourth embodiments above can be at least partially implemented as a computer program module, which can perform corresponding functions when the computer program module is run.

[0098] The fifth exemplary embodiment of this disclosure provides an electronic device.

[0099] Figure 7 A schematic block diagram of an electronic device provided in an embodiment of the present disclosure is shown.

[0100] Reference Figure 7 As shown, the electronic device 700 provided in this embodiment includes a processor 701, a communication interface 702, a memory 703, and a communication bus 704. The processor 701, the communication interface 702, and the memory 703 communicate with each other through the communication bus 704. The memory 703 is used to store computer programs. When the processor 701 executes the program stored in the memory, it implements the method for constructing an initial arrival picking model or the initial arrival picking method as described above.

[0101] A sixth exemplary embodiment of this disclosure also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method for constructing an initial arrival picking model or the initial arrival picking method as described above.

[0102] The computer-readable storage medium may be included in the device or apparatus described in the above embodiments; or it may exist independently and not assembled into the device or apparatus. The computer-readable storage medium carries one or more programs that, when executed, implement the method according to the embodiments of this disclosure.

[0103] According to embodiments of this disclosure, the computer-readable storage medium can be a non-volatile computer-readable storage medium, such as including, but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this disclosure, the computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0104] It should be noted that the collection, gathering, updating, analysis, processing, use, transmission, and storage of user personal information involved in the technical solutions provided in this disclosure comply with the provisions of relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. Necessary measures are taken to prevent unauthorized access to user personal information data and to safeguard user personal information security, network security, and national security.

[0105] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0106] The above description is merely a specific embodiment of this disclosure, enabling those skilled in the art to understand or implement it. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for constructing an initial arrival picking model, characterized in that, include: For the initial arrival low signal-to-noise ratio area of ​​the target exploration area, multiple seismic sampling data corresponding to the location of each target sampling point are obtained from multiple single-shot excitations; The multiple seismic sampling data are vertically stacked to obtain stacked seismic data, and the first arrival of the target is obtained by picking the first arrival of the stacked seismic data. Seismic sampling data corresponding to multiple target sampling point locations are used as training inputs to an artificial intelligence model. The artificial intelligence model outputs the first arrival picking results. The target first arrivals are used as training labels to supervise the training of the artificial intelligence model. The trained artificial intelligence model is used as the first arrival picking model.

2. The method according to claim 1, characterized in that, The multiple seismic sampling data were acquired through the following methods: Based on the existing seismic data of the target exploration area, candidate sampling points corresponding to different noise types in the initial arrival low signal-to-noise ratio area are determined; Based on the location of the candidate sampling points, the existing first-arrival signal-to-noise ratio information, and the preset coverage information of the noise type, the location of the target sampling point for performing one-point multi-sampling and the corresponding number of samplings are determined. At the target sampling point location within the low signal-to-noise ratio region, multiple single-shot excitation results are collected according to the specified number of acquisitions to obtain multiple seismic sampling data.

3. The method according to claim 2, characterized in that, The preset coverage information is used to indicate the preset quantity or preset proportion of noise types to be covered by the training samples; Based on the location of the candidate sampling points, the existing first-arrival signal-to-noise ratio information, and the preset coverage information of the noise type, the target sampling point location and the corresponding number of samplings for one-point multi-sampling are determined, including: Based on the preset coverage information, the positions of candidate sampling points corresponding to all or part of the noise types are determined as the target sampling point positions for one-point multi-sampling. Based on the existing first-to-first signal-to-noise ratio (SNR) information of the target sampling point location, determine the minimum number of stacking operations required to satisfy the condition that the first-to-first SNR exceeds the threshold value after vertical stacking; the threshold value is the division threshold corresponding to the first-to-first low SNR region and the first-to-first high SNR region. Randomly generate a number of sampling attempts that satisfy the minimum number of superpositions, or use an input number that exceeds the minimum number of superpositions as the number of sampling attempts.

4. The method according to claim 3, characterized in that, The number of times each target sampling point is superimposed satisfies the following expression: Among them, S p / N p S / N represents the initial-to-noise ratio after vertical stacking; S / N represents the existing initial-to-noise ratio corresponding to the current target sampling point position; P represents the number of stacking times; C represents the boundary value corresponding to the division between the initial-to-low signal-to-noise ratio region and the initial-to-high signal-to-noise ratio region. By solving the above expression, the minimum value of P is obtained as the minimum number of superpositions.

5. The method according to claim 1, characterized in that, Also includes: For the target exploration area, the area is divided into a low signal-to-noise ratio zone and a high signal-to-noise ratio zone. The regions from initial low signal-to-noise ratio (SNR) and from initial high SNR are divided as follows: Based on the existing seismic data of the target exploration area, determine the total energy of the first arrival wave and the total noise energy; The overall signal-to-noise ratio is obtained by calculating the ratio of the total energy of the first arrival wave to the total energy of the noise. Based on the overall signal-to-noise ratio, a boundary value is determined; the boundary value is the division threshold corresponding to the initial low signal-to-noise ratio region and the initial high signal-to-noise ratio region. Based on the boundary value and the regional signal-to-noise ratio of each region, the regions are divided into regions with low initial signal-to-noise ratio and regions with high initial signal-to-noise ratio.

6. A method for initial arrival pickup, characterized in that, include: Acquire seismic data to be processed; The seismic data to be processed is input into a pre-constructed first-arrival picking model, and the corresponding first-arrival picking result is output; wherein, the first-arrival picking model is constructed using the method described in any one of claims 1-5.

7. An apparatus for constructing an initial arrival picking model, characterized in that, include: The data acquisition module is used to acquire multiple seismic sampling data obtained from multiple single-shot excitations for each target sampling point location in the initial low signal-to-noise ratio area of ​​the target exploration area. The training label calculation module is used to vertically overlay the multiple seismic sampling data to obtain overlaid seismic data, and to pick the first arrival of the target from the overlaid seismic data; The training module is used to input seismic sampling data corresponding to multiple target sampling point locations as training input to an artificial intelligence model. The artificial intelligence model outputs the first arrival picking result, and the target first arrival is used as a training label to supervise the training of the artificial intelligence model. The trained artificial intelligence model is used as the first arrival picking model.

8. A device for initial pickup, characterized in that, include: The data acquisition module is used to acquire the seismic data to be processed. The data processing module is used to input the seismic data to be processed into a pre-constructed first arrival picking model and output the corresponding first arrival picking result; wherein the first arrival picking model is constructed by the method of any one of claims 1-5 or by the device of claim 7.

9. An electronic device, characterized in that, It includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, when executing a program stored in memory, implements the method of any one of claims 1-6.

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

Images