A method and device for identifying and suppressing multiple waves of seafloor seismograph data

Abstract

The present invention relates to the field of marine geophysical exploration technology, and provides a method and device for identifying and suppressing multiple waves of seafloor seismograph data, the method comprising: obtaining simulated seafloor seismograph data and actual seafloor seismograph data collected by a field OBS station; inputting actual single-shot seismic data into a trained support vector machine to obtain actual single-shot seismic data with multiple waves and actual single-shot seismic data without multiple waves; inputting the actual single-shot seismic data with multiple waves into a trained convolutional neural network model to obtain the multiple wave period of the actual single-shot seismic data with multiple waves. The present invention can be unaffected by the type of multiple waves for active source low-frequency OBS records with long survey lines and great depths, and only relies on periodicity, efficiently confirms and searches for multiple reflection / refracted waves of different periods and different types, and suppresses various types of periodic multiple waves according to multiple determined prediction step sizes.

Description

Technical Field

[0001] The invention relates to the technical field of marine geophysical exploration, and in particular to a method and device for identifying and suppressing multiple waves of seafloor seismograph data. Background Art

[0002] Ocean bottom seismometer (OBS) exploration is a type of detection method with large offset and wide azimuth. The active source is usually equipped with an airgun source, and the OBS receives reflected waves and refracted waves on the seabed. In seabed seismometer exploration, the OBS stations are far apart, and the length of the OBS observation line of a single station is more than 100 km. The data is rich in complex and changeable multiple reflection waves. However, since the seawater depth covered by the OBS survey line usually has a certain degree of variation, the multiple reflection / refracted waves of the seabed passing through different depths of seawater and different paths do not necessarily show a regular continuous relationship on the OBS common detection point gather. Therefore, although the multiple waves on each shot channel in the OBS data are periodic, the apparent period between different shots on the same detection point gather is complex. In addition, the signal frequency in OBS records is relatively low, and its effective frequency band in regional structural detection is basically no more than 20 Hz. Because the sea depth is relatively shallow, the seabed multiple reflection / refracted waves (or full-range multiple waves) often appear in the form of accompanying waves immediately following the seabed primary wave, and it is difficult to separate from the primary wave, and it affects the waveform of the primary wave from the subsequent strata. In an actual OBS seismic profile, there is usually not only one type of multiple waves. In most cases, there will be multiple types of multiple wave groups originating from different interfaces, and the various types of multiple waves in the gun channel show certain periodic characteristics.

[0003] In the prediction / suppression of multiple waves in seafloor seismograph detection data, there are currently multiple wave prediction methods based on Kirchhoff integrals and multiple wave suppression methods based on sparse inversion, all of which are suitable for processing primary reflection waves in observation systems with small detection depths, short survey line coverage distances, and relatively dense OBS station distribution. For OBS observation systems serving regional geological surveys, the detection depth exceeds 30km, the single station recording coverage distance exceeds 100km, and the OBS station distribution is even more sparse. The effect of suppressing multiple waves in such OBS station data using the above methods is not ideal.

[0004] To this end, the present invention provides a method and device for identifying and suppressing multiple waves in seafloor seismograph data, which utilizes the periodicity of multiple waves, automatically detects the time-varying period of multiple reflection / refracting waves on the seafloor in active source OBS data, and performs predictive deconvolution suppression of multiple reflection / refracting waves. Summary of the invention

[0005] In view of the problems existing in the prior art, the present invention provides a method and device for identifying and suppressing multiple waves in seafloor seismograph data, so as to solve the problem that the existing multiple wave suppression method is not suitable for suppressing seafloor multiple waves in active source seismic data in an observation system with a large detection depth, a long survey line coverage distance, and a relatively sparse distribution of OBS stations.

[0006] In a first aspect, the present invention provides a method for identifying and suppressing multiple waves in seafloor seismograph data, the method comprising:

[0007] Step S101: Establish an underground medium velocity model, set up an OBS observation system, perform earthquake forward modeling, and obtain simulated seafloor seismograph data;

[0008] Step S102: Acquire actual seafloor seismograph data collected by a field OBS station;

[0009] Step S103: performing minimum phase wavelet shaping on the common detection point gathers of the simulated seafloor seismograph data and the common detection point gathers of the actual seafloor seismograph data, respectively, to obtain the simulated seafloor seismograph data and the actual seafloor seismograph data satisfying the minimum phase wavelet condition;

[0010] Step S104: using the autocorrelation function, autocorrelation processing is performed on the simulated seafloor seismograph data and the actual seafloor seismograph data that meet the minimum phase wavelet condition, to obtain an autocorrelation data set of common detection point gathers of the simulated seafloor seismograph data and an autocorrelation data set of common detection point gathers of the actual seafloor seismograph data; autocorrelation data of the simulated single-shot seismic data and autocorrelation data of the actual single-shot seismic data are obtained according to the autocorrelation data set of common detection point gathers of the simulated seafloor seismograph data and the autocorrelation data set of common detection point gathers of the actual seafloor seismograph data;

[0011] Step S105: extracting feature data of the autocorrelation data of the simulated single-shot seismic data and the autocorrelation data of the actual single-shot seismic data respectively, and using the feature data to train a support vector machine model to obtain a trained support vector machine model;

[0012] Step S106: using the autocorrelation data of the simulated single shot seismic data and the autocorrelation data of the actual single shot seismic data to train a convolutional neural network model, to obtain a trained convolutional neural network model;

[0013] Step S107: inputting the actual single shot seismic data into the trained support vector machine model to obtain the actual single shot seismic data with multiple waves and the actual single shot seismic data without multiple waves;

[0014] Step S108: inputting the actual single-shot seismic data with multiple waves into the trained convolutional neural network model to obtain the multiple wave periods of the actual single-shot seismic data with multiple waves; taking all the multiple wave periods of the actual single-shot seismic data with multiple waves as prediction step lengths, testing and determining prediction deconvolution parameters, suppressing multiple waves by using prediction deconvolution according to the determined prediction deconvolution parameters, and obtaining the actual single-shot seismic data after the multiple waves are suppressed;

[0015] Step S109: using the autocorrelation function, obtaining the autocorrelation data of the actual single-shot seismic data after multiple wave suppression;

[0016] Step S110: replacing the autocorrelation data of the actual single-shot seismic data in step S105 and step S106 with the autocorrelation data of the actual single-shot seismic data after multiple wave suppression, replacing the actual single-shot seismic data in step S107 with the actual single-shot seismic data after multiple wave suppression, repeating steps S105 to S109 until the actual single-shot seismic data after multiple wave suppression is input into the trained support vector machine model in step S107, and only the actual single-shot seismic data without multiple waves is obtained;

[0017] Step S111: Execute steps S105 to S110 on all actual single-shot seismic data on the common detection point gather of the actual seafloor seismograph data to obtain all actual single-shot seismic data without multiple waves.

[0018] Furthermore, the step S103 further comprises: before performing minimum phase wavelet shaping on the common detection point gathers of the actual seafloor seismograph data, preprocessing the actual seafloor seismograph data, wherein the preprocessing comprises bandpass filtering, random noise removal and normalization processing.

[0019] Furthermore, in step S105, the characteristic data includes peak position, peak amplitude, peak interval, decay speed and symmetry index.

[0020] Furthermore, the step S105 includes:

[0021] Step S1051: Divide the characteristic data into a first training set and a first test set, wherein the ratio of characteristic data of autocorrelation data of simulated single shot seismic data to characteristic data of autocorrelation data of actual single shot seismic data in the first training set is 3:1, and the first test set is characteristic data of autocorrelation data of simulated single shot seismic data;

[0022] Step S1052: Mark the first training set as a training set of feature data with multiple waves and feature data without multiple waves to train a support vector machine model, learn the boundary between the periodic characteristics and the non-periodic characteristics of the signal, and obtain a trained support vector machine model; use the first test set to evaluate the performance of the trained support vector machine model, and when the performance meets the standard, obtain a trained support vector machine model.

[0023] Furthermore, the step S106 includes:

[0024] Step S1061: Divide the autocorrelation data of simulated single-shot seismic data and the autocorrelation data of actual single-shot seismic data into a second training set and a second test set, wherein the ratio of the autocorrelation data of simulated single-shot seismic data to the autocorrelation data of actual single-shot seismic data in the second training set is 3:1, and the second test set is the autocorrelation data of simulated single-shot seismic data;

[0025] Step S1062: input the second training set into the input layer, use multiple convolutional layers with a step size of 2 to extract local features in the autocorrelation data, use a fully connected layer to integrate the local features and predict the delay time of the sub-peak, output the predicted delay time of the sub-peak in the output layer, and obtain the trained convolutional neural network model, wherein the predicted delay time of the sub-peak is a multiple wave period; use the second test set data to evaluate the performance of the convolutional neural network model, and when the performance meets the standard, obtain the trained convolutional neural network model.

[0026] In a second aspect, an embodiment of the present application provides a device for identifying and suppressing multiple waves of seafloor seismograph data, the device comprising:

[0027] Forward simulation module: establish underground medium velocity model, set up OBS observation system, conduct earthquake forward simulation, and obtain simulated seafloor seismograph data;

[0028] Data acquisition module: obtains actual seafloor seismograph data collected by field OBS stations;

[0029] Minimum phase wavelet shaping module: performing minimum phase wavelet shaping on the common detection point gathers of the simulated seafloor seismograph data and the common detection point gathers of the actual seafloor seismograph data, respectively, to obtain the simulated seafloor seismograph data and the actual seafloor seismograph data that meet the minimum phase wavelet condition;

[0030] The first autocorrelation module: using the autocorrelation function, autocorrelation processing is performed on the simulated seafloor seismograph data and the actual seafloor seismograph data that meet the minimum phase wavelet condition, to obtain the autocorrelation data set of the common detection point gather of the simulated seafloor seismograph data and the autocorrelation data set of the common detection point gather of the actual seafloor seismograph data; the autocorrelation data of the simulated single-shot seismic data and the autocorrelation data of the actual single-shot seismic data are obtained according to the autocorrelation data set of the common detection point gather of the simulated seafloor seismograph data and the autocorrelation data set of the common detection point gather of the actual seafloor seismograph data;

[0031] Support vector machine model training module: extracting feature data of autocorrelation data of simulated single shot seismic data and autocorrelation data of actual single shot seismic data respectively, using the feature data to train a support vector machine model, and obtaining a trained support vector machine model;

[0032] Convolutional neural network model training module: using the autocorrelation data of simulated single-shot seismic data and the autocorrelation data of actual single-shot seismic data to train the convolutional neural network model and obtain a trained convolutional neural network model;

[0033] Multiple wave identification module: input the actual single shot seismic data into the trained support vector machine model to obtain the actual single shot seismic data with multiple waves and the actual single shot seismic data without multiple waves;

[0034] Multiple wave suppression module: input the actual single shot seismic data with multiple waves into the trained convolutional neural network model to obtain the multiple wave periods of the actual single shot seismic data with multiple waves; use all the multiple wave periods of the actual single shot seismic data with multiple waves as prediction step lengths, test and determine prediction deconvolution parameters, suppress multiple waves by using prediction deconvolution according to the determined prediction deconvolution parameters, and obtain the actual single shot seismic data after the multiple waves are suppressed;

[0035] The second autocorrelation module: using the autocorrelation function, obtains the autocorrelation data of the actual single-shot seismic data after multiple wave suppression;

[0036] Loop module: replace the autocorrelation data of the actual single-shot seismic data in the support vector machine model training module and the convolutional neural network model training module with the autocorrelation data of the actual single-shot seismic data after multiple wave suppression, replace the actual single-shot seismic data in the multiple wave identification module with the actual single-shot seismic data after multiple wave suppression, and repeatedly run the support vector machine model training module, the convolutional neural network model training module, the multiple wave identification module, the multiple wave suppression module and the second autocorrelation module until the actual single-shot seismic data after multiple wave suppression is input into the trained support vector machine model in the multiple wave identification module, and only the actual single-shot seismic data without multiple waves is obtained;

[0037] Repeated execution module: execute support vector machine model training module, convolutional neural network model training module, multiple wave identification module, multiple wave suppression module, second autocorrelation module and loop module for all single-shot seismic data on the common detection point gather of the actual seafloor seismometer data to obtain all actual single-shot seismic data without multiple waves.

[0038] In a third aspect, the present invention provides an electronic device, comprising:

[0039] processor;

[0040] Memory;

[0041] And a computer program, wherein the computer program is stored in the memory, and the computer program includes instructions, and when the instructions are executed by the processor, the electronic device executes any one of the methods described in the first aspect.

[0042] In a fourth aspect, the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the methods described in the first aspect.

[0043] Based on the above invention content, compared with the prior art, the present invention utilizes the periodicity of multiple waves, automatically detects and suppresses the time-varying period of multiple reflection / refracted waves on the seabed of active source OBS data, and carries out the prediction deconvolution suppression of multiple reflection / refracted waves. For active source low-frequency OBS records with long survey lines and large depths, it can be unaffected by the type of multiple waves, and only rely on periodicity to efficiently confirm and search for multiple reflection / refracted waves of different periods and types, and suppress various types of periodic multiple waves according to multiple determined prediction step sizes.

[0044] The following technical effects are achieved:

[0045] (1) The only raw data required for deconvolution operation is the OBS seismic channel itself. There are no more requirements for the parameters of the layered structure including the seawater layer, which facilitates the specific implementation of the operation even in the complex seabed and the strata below. For example, the prior art requires the "depth" of the seawater layer to be known in advance to calculate the prediction step length; while the prediction step length in the present invention is obtained by learning the convolutional neural network model in the autocorrelation records with multiple waves identified by the support vector machine model, so there is no need to obtain the "depth" of the seawater layer in advance in the present invention.

[0046] (2) The present invention implements minimum phase wavelet shaping, making it possible to predict deconvolution and eliminate multiple waves under stereo airgun excitation conditions.

[0047] (3) It is crucial to obtain the prediction step size in the process of predictive deconvolution of active source OBS seismic data. In a small-volume, low signal-to-noise ratio autocorrelation data set, the present invention uses the autocorrelation function and combines the support vector machine model to accurately determine whether there are multiple waves. The convolutional neural network model ensures the accuracy and stability of the delay time. Compared with the classical automatic picking method, it has higher accuracy, faster speed, and is suitable for more concealed data bodies.

[0048] (4) The predictive deconvolution used in the calculation process of the present invention has few and simple implementation conditions, and is suitable for suppressing periodic multiple waves. It can suppress multiple waves of different periods in the same channel data, and there is no need to determine the type of multiple waves. In the process of predictive deconvolution of active source OBS seismic data, automatic search, automatic identification, and simultaneous suppression of periodic multiple reflection waves and refraction waves are realized. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0050] Figure 1 It is a flow chart of a method for identifying and suppressing multiple waves of seafloor seismograph data provided by an embodiment of the present application;

[0051] Figure 2 A comparison chart of the converted display records of actual seafloor seismograph data before and after being processed by the method of the present application provided in an embodiment of the present application;

[0052] Figure 3 A comparison diagram of the autocorrelation profile of actual seismic data of common receiving point gathers before and after multiple wave suppression provided in an embodiment of the present application;

[0053] Figure 4 It is a structural block diagram of a device for identifying and suppressing multiple waves of seafloor seismograph data provided by an embodiment of the present application;

[0054] Figure 5 It is a structural schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0055] In order to better understand the technical solution of the present invention, the embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0056] It should be clear that the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0057] See also Figure 1 , is a flow chart of a method for identifying and suppressing multiple waves of seafloor seismograph data provided by an embodiment of the present invention. Figure 1 As shown, the method specifically includes:

[0058] Step S101: Establish an underground medium velocity model, set up an OBS observation system, perform earthquake forward modeling, and obtain simulated seafloor seismograph data.

[0059] Step S102: Acquire actual seafloor seismograph data collected by a field OBS station.

[0060] Step S103: performing minimum phase wavelet shaping on the common detection point gathers of the simulated seafloor seismograph data and the common detection point gathers of the actual seafloor seismograph data respectively to obtain simulated seafloor seismograph data and actual seafloor seismograph data satisfying minimum phase wavelet conditions.

[0061] Before performing minimum phase wavelet shaping on the common detection point gathers of the actual seafloor seismograph data, the actual seafloor seismograph data is preprocessed, and the preprocessing includes bandpass filtering, random noise removal and normalization processing.

[0062] Step S104: using the autocorrelation function, autocorrelation processing is performed on the simulated seafloor seismograph data and the actual seafloor seismograph data that meet the minimum phase wavelet condition, to obtain an autocorrelation data set of common-detection point gathers of the simulated seafloor seismograph data and an autocorrelation data set of common-detection point gathers of the actual seafloor seismograph data; based on the autocorrelation data set of common-detection point gathers of the simulated seafloor seismograph data and the autocorrelation data set of common-detection point gathers of the actual seafloor seismograph data, the autocorrelation data of the simulated single-shot seismic data and the autocorrelation data of the actual single-shot seismic data are obtained.

[0063] Step S105: extracting feature data of the autocorrelation data of the simulated single-shot seismic data and the autocorrelation data of the actual single-shot seismic data respectively, and using the feature data to train a support vector machine model to obtain a trained support vector machine model.

[0064] The characteristic data includes peak position, peak amplitude, peak spacing, decay speed and symmetry index. The step S105 includes:

[0065] Step S1051: Divide the characteristic data into a first training set and a first test set, wherein the ratio of characteristic data of autocorrelation data of simulated single shot seismic data to characteristic data of autocorrelation data of actual single shot seismic data in the first training set is 3:1, and the first test set is characteristic data of autocorrelation data of simulated single shot seismic data;

[0066] Step S1052: Mark the first training set as a training set of feature data with multiple waves and feature data without multiple waves to train a support vector machine model, learn the boundary between the periodic characteristics and the non-periodic characteristics of the signal, and obtain a trained support vector machine model; use the first test set to evaluate the performance of the trained support vector machine model, and when the performance meets the standard, obtain a trained support vector machine model.

[0067] Step S106: using the autocorrelation data of the simulated single shot seismic data and the autocorrelation data of the actual single shot seismic data, training the convolutional neural network model to obtain a trained convolutional neural network model. The step S106 includes:

[0068] Step S1061: Divide the autocorrelation data of simulated single-shot seismic data and the autocorrelation data of actual single-shot seismic data into a second training set and a second test set, wherein the ratio of the autocorrelation data of simulated single-shot seismic data to the autocorrelation data of actual single-shot seismic data in the second training set is 3:1, and the second test set is the autocorrelation data of simulated single-shot seismic data;

[0069] Step S1062: input the second training set into the input layer, use multiple convolutional layers with a step size of 2 to extract local features in the autocorrelation data, use a fully connected layer to integrate the local features and predict the delay time of the sub-peak, output the predicted delay time of the sub-peak in the output layer, and obtain the trained convolutional neural network model, wherein the predicted delay time of the sub-peak is a multiple wave period; use the second test set data to evaluate the performance of the convolutional neural network model, and when the performance meets the standard, obtain the trained convolutional neural network model.

[0070] Step S107: input the actual single-shot seismic data into the trained support vector machine model to obtain the actual single-shot seismic data with multiple waves and the actual single-shot seismic data without multiple waves.

[0071] Step S108: inputting the actual single-shot seismic data with multiple waves into the trained convolutional neural network model to obtain the multiple wave periods of the actual single-shot seismic data with multiple waves; taking all the multiple wave periods of the actual single-shot seismic data with multiple waves as prediction step lengths, testing and determining prediction deconvolution parameters, suppressing the multiple waves by using prediction deconvolution according to the determined prediction deconvolution parameters, and obtaining the actual single-shot seismic data after the multiple waves are suppressed.

[0072] Step S109: using the autocorrelation function, obtaining the autocorrelation data of the actual single-shot seismic data after multiple wave suppression.

[0073] Step S110: Replace the autocorrelation data of the actual single-shot seismic data in step S105 and step S106 with the autocorrelation data of the actual single-shot seismic data after multiple wave suppression, replace the actual single-shot seismic data in step S107 with the actual single-shot seismic data after multiple wave suppression, repeat steps S105-S109 until the actual single-shot seismic data after multiple wave suppression is input into the trained support vector machine model in step S107, and only the actual single-shot seismic data without multiple waves is obtained.

[0074] Step S111: Execute steps S105 to S110 on all actual single-shot seismic data on the common detection point gather of the actual seafloor seismograph data to obtain all actual single-shot seismic data without multiple waves.

[0075] To give an example, in this embodiment, the method of the present application is further illustrated by processing actual seafloor seismograph data collected at the K01 station in the South Yellow Sea.

[0076] Figure 2 This is a comparison chart of the converted display records of actual seafloor seismograph data before and after processing using the method of this application. Figure 2 (a) is the converted display record of the actual seafloor seismograph data collected at station K01 in the South Yellow Sea; Figure 2 Middle (b) is the converted display record after preprocessing the actual seafloor seismograph data collected at station K01 in the South Yellow Sea; Figure 2 (c) is the converted display record after the actual seafloor seismograph data collected by the K01 station in the South Yellow Sea was processed by this method. Figure 2 (a) and Figure 2 As can be seen in (b), after preprocessing, the phase axis of the actual seafloor seismograph data was identified, but there was strong interference from multiple waves. Due to the influence of the seafloor multiple waves, the characteristics of the seismic wave group in the profile are not obvious. The combined influence of the primary wave and the seafloor multiple waves makes the seismic reflection of the deep target layer often have 3 to 4 high-energy peaks and troughs, and the energy between them is comparable, which brings great difficulty to the identification and interpretation of the seismic phase. By comparing Figure 2 (b) and Figure 2 It can be seen from the arrow position in the middle (c) that after being processed by the method of the present application, the multiple waves are well suppressed, the seismic wave group characteristics are significantly improved, multiple peaks and troughs with equivalent energy are suppressed, the event axis becomes relatively simple, and the entire profile is relatively clean, highlighting the main event axis information and wave group characteristics, which is more conducive to the subsequent identification and interpretation of the target layer.

[0077] Extract all actual single-shot seismic data according to the common receiving points to obtain the actual seismic data of the common receiving point gather. Figure 3 This is a comparison diagram of the autocorrelation profile of the actual seismic data of the common receiving point gather before and after multiple wave suppression; Figure 3 (a) is the autocorrelation profile of the actual seismic data collected from the common receiver gather at the K01 station in the South Yellow Sea; Figure 3 (b) is the autocorrelation profile of the actual seismic data of the common receiving point gather obtained after 6 multiple wave suppressions; Figure 3 (c) is the autocorrelation profile of the actual seismic data of the common receiving point gather without multiple waves. Figure 3 It can be seen from (a) to (c) that for the actual seafloor seismograph data collected at the K01 station in the South Yellow Sea, after six times of multiple wave suppression, there are only a small number of secondary extreme values ​​near the autocorrelation principal axis. The autocorrelation profile of the actual seismic data without multiple waves finally obtained only has the autocorrelation principal axis, the secondary extreme values ​​disappear, and the energy is concentrated, indicating that the method of the present application effectively identifies and suppresses multiple waves.

[0078] The present invention utilizes the periodicity of multiple waves, and automatically detects and suppresses the time-varying period of multiple reflection / refracted waves of active source OBS data by training support vector machine models and convolutional neural network models, and carries out predictive deconvolution suppression of multiple reflection / refracted waves. For active source low-frequency OBS data with long survey lines and large depths, it can be unaffected by the type of multiple waves, relying only on periodicity, efficiently searching and confirming multiple reflection / refracted waves of different periods and types, and suppressing various types of periodic multiple waves according to multiple determined prediction step sizes. From the above description, it can be seen that the above-mentioned embodiments of the present application achieve the following technical effects:

[0079] (1) The only raw data required for deconvolution operation is the OBS seismic channel itself. There are no more requirements for the parameters of the layered structure including the seawater layer, which facilitates the specific implementation of the operation even in the complex seabed and the strata below. For example, the prior art requires the "depth" of the seawater layer to be known in advance to calculate the prediction step length; while the prediction step length in the present invention is obtained by learning the convolutional neural network model in the autocorrelation records with multiple waves identified by the support vector machine model, so there is no need to obtain the "depth" of the seawater layer in advance in the present invention.

[0080] (2) The present invention implements minimum phase wavelet shaping, making it possible to predict deconvolution and eliminate multiple waves under stereo airgun excitation conditions.

[0081] (3) It is crucial to obtain the prediction step size in the process of predictive deconvolution of active source OBS seismic data. In a small-volume, low signal-to-noise ratio autocorrelation data set, the present invention uses the autocorrelation function and combines the support vector machine model to accurately determine whether there are multiple waves. The convolutional neural network model ensures the accuracy and stability of the delay time. Compared with the classical automatic picking method, it has higher accuracy, faster speed, and is suitable for more concealed data bodies.

[0082] (4) The predictive deconvolution used in the calculation process of the present invention has few and simple implementation conditions, and is suitable for suppressing periodic multiple waves. It can suppress multiple waves of different periods in the same channel data, and there is no need to determine the type of multiple waves. In the process of predictive deconvolution of active source OBS seismic data, automatic search, automatic identification, and simultaneous suppression of periodic multiple reflection waves and refraction waves are realized.

[0083] Corresponding to the above-mentioned embodiment, the present invention also provides a device for identifying and suppressing multiple waves of seafloor seismograph data.

[0084] See also Figure 4 , is a structural block diagram of a device for identifying and suppressing multiple waves of seafloor seismograph data provided by an embodiment of the present invention. Figure 4 As shown, it mainly includes the following modules.

[0085] Forward simulation module 401: establish underground medium velocity model, set up OBS observation system, perform earthquake forward simulation, and obtain simulated seafloor seismograph data;

[0086] Data acquisition module 402: acquiring actual seafloor seismograph data collected by field OBS stations;

[0087] Minimum phase wavelet shaping module 403: performing minimum phase wavelet shaping on the common detection point gathers of the simulated seafloor seismograph data and the common detection point gathers of the actual seafloor seismograph data, respectively, to obtain the simulated seafloor seismograph data and the actual seafloor seismograph data satisfying the minimum phase wavelet condition;

[0088] The first autocorrelation module 404: using the autocorrelation function, autocorrelation processing is performed on the simulated seafloor seismograph data and the actual seafloor seismograph data that meet the minimum phase wavelet condition, to obtain an autocorrelation data set of common detection point gathers of the simulated seafloor seismograph data and an autocorrelation data set of common detection point gathers of the actual seafloor seismograph data; autocorrelation data of the simulated single-shot seismic data and autocorrelation data of the actual single-shot seismic data are obtained according to the autocorrelation data set of common detection point gathers of the simulated seafloor seismograph data and the autocorrelation data set of common detection point gathers of the actual seafloor seismograph data;

[0089] Support vector machine model training module 405: extracting feature data of the autocorrelation data of simulated single shot seismic data and the autocorrelation data of actual single shot seismic data respectively, and using the feature data to train a support vector machine model to obtain a trained support vector machine model;

[0090] Convolutional neural network model training module 406: using the autocorrelation data of simulated single shot seismic data and the autocorrelation data of actual single shot seismic data to train the convolutional neural network model and obtain a trained convolutional neural network model;

[0091] Multiple wave identification module 407: inputting the actual single shot seismic data into the trained support vector machine model to obtain the actual single shot seismic data with multiple waves and the actual single shot seismic data without multiple waves;

[0092] The multiple wave suppression module 408: inputs the actual single shot seismic data with multiple waves into the trained convolutional neural network model to obtain the multiple wave periods of the actual single shot seismic data with multiple waves; uses all the multiple wave periods of the actual single shot seismic data with multiple waves as prediction step lengths, tests and determines prediction deconvolution parameters, suppresses multiple waves by using prediction deconvolution according to the determined prediction deconvolution parameters, and obtains the actual single shot seismic data after the multiple waves are suppressed;

[0093] The second autocorrelation module 409: using the autocorrelation function, obtains the autocorrelation data of the actual single shot seismic data after multiple wave suppression;

[0094] Loop module 410: replace the autocorrelation data of the actual single-shot seismic data in the support vector machine model training module and the convolutional neural network model training module with the autocorrelation data of the actual single-shot seismic data after multiple wave suppression, replace the actual single-shot seismic data in the multiple wave identification module with the actual single-shot seismic data after multiple wave suppression, and repeatedly run the support vector machine model training module, the convolutional neural network model training module, the multiple wave identification module, the multiple wave suppression module and the second autocorrelation module until the actual single-shot seismic data after multiple wave suppression is input into the trained support vector machine model in the multiple wave identification module, and only the actual single-shot seismic data without multiple waves is obtained;

[0095] Repeat the execution of module 411: execute the support vector machine model training module, the convolutional neural network model training module, the multiple wave identification module, the multiple wave suppression module, the second autocorrelation module and the loop module for all single-shot seismic data on the common detection point gather of the actual seafloor seismometer data to obtain all the actual single-shot seismic data without multiple waves.

[0096] It should be pointed out that the specific contents involved in the embodiments of the present invention can be found in the description of the above method embodiments, and for the sake of brevity, they will not be repeated here.

[0097] Corresponding to the above embodiment, an embodiment of the present invention further provides an electronic device.

[0098] See also Figure 5 , is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. Figure 5 As shown, the electronic device 500 may include: a processor 501, a memory 502 and a communication unit 503. These components communicate via one or more buses. Those skilled in the art will appreciate that the electronic device structure shown in the figure does not limit the embodiments of the present invention. It may be a bus structure or a star structure, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0099] The communication unit 503 is used to establish a communication channel so that the electronic device can communicate with other devices.

[0100] The processor 501 is the control center of the electronic device. It uses various interfaces and lines to connect various parts of the entire electronic device. It runs or executes software programs and / or modules stored in the memory 502, and calls data stored in the memory to perform various functions of the electronic device and / or process data. The processor can be composed of an integrated circuit (IC), for example, it can be composed of a single packaged IC, or it can be composed of a plurality of packaged ICs with the same or different functions. For example, the processor 501 can include only a central processing unit (CPU). In an embodiment of the present invention, the CPU can be a single computing core or multiple computing cores.

[0101] The memory 502 is used to store the execution instructions of the processor 501. The memory 502 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0102] When the execution instructions in the memory 502 are executed by the processor 501 , the electronic device 500 is enabled to execute part or all of the steps in the above method embodiment.

[0103] Corresponding to the above embodiment, the embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium may store a program, wherein when the program is running, the device where the computer-readable storage medium is located may be controlled to execute some or all of the steps in the above method embodiment. In a specific implementation, the computer-readable storage medium may be a magnetic disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM).

[0104] Corresponding to the above embodiment, an embodiment of the present invention further provides a computer program product, which includes executable instructions. When the executable instructions are executed on a computer, the computer executes part or all of the steps in the above method embodiment.

[0105] In the embodiments of the present invention, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. A and B may be singular or plural. The character " / " generally indicates that the previous and subsequent associated objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b and c can be represented by: a, b, c, ab, ac, bc or abc, where a, b, c may be single or multiple.

[0106] Those skilled in the art will appreciate that the various units and algorithm steps described in the embodiments disclosed herein can be implemented in a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0107] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0108] In several embodiments provided by the present invention, if any function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0109] The above description is only a specific implementation of the present invention. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention.

Claims

1. A method for identifying and suppressing multiple waves of seafloor seismograph data, characterized in that: The method comprises: Step S101: Establish an underground medium velocity model, set up an OBS observation system, perform earthquake forward modeling, and obtain simulated seafloor seismograph data; Step S102: Acquire actual seafloor seismograph data collected by a field OBS station; Step S103: performing minimum phase wavelet shaping on the common detection point gathers of the simulated seafloor seismograph data and the common detection point gathers of the actual seafloor seismograph data, respectively, to obtain the simulated seafloor seismograph data and the actual seafloor seismograph data satisfying the minimum phase wavelet condition; Step S104: using the autocorrelation function, autocorrelation processing is performed on the simulated seafloor seismograph data and the actual seafloor seismograph data that meet the minimum phase wavelet condition, to obtain an autocorrelation data set of common detection point gathers of the simulated seafloor seismograph data and an autocorrelation data set of common detection point gathers of the actual seafloor seismograph data; autocorrelation data of the simulated single-shot seismic data and autocorrelation data of the actual single-shot seismic data are obtained according to the autocorrelation data set of common detection point gathers of the simulated seafloor seismograph data and the autocorrelation data set of common detection point gathers of the actual seafloor seismograph data; Step S105: extracting feature data of the autocorrelation data of the simulated single-shot seismic data and the autocorrelation data of the actual single-shot seismic data respectively, and using the feature data to train a support vector machine model to obtain a trained support vector machine model; Step S106: using the autocorrelation data of the simulated single shot seismic data and the autocorrelation data of the actual single shot seismic data to train a convolutional neural network model, to obtain a trained convolutional neural network model; Step S107: inputting the actual single shot seismic data into the trained support vector machine model to obtain the actual single shot seismic data with multiple waves and the actual single shot seismic data without multiple waves; Step S108: inputting the actual single-shot seismic data with multiple waves into the trained convolutional neural network model to obtain the multiple wave periods of the actual single-shot seismic data with multiple waves; taking all the multiple wave periods of the actual single-shot seismic data with multiple waves as prediction step lengths, testing and determining prediction deconvolution parameters, suppressing multiple waves by using prediction deconvolution according to the determined prediction deconvolution parameters, and obtaining the actual single-shot seismic data after the multiple waves are suppressed; Step S109: using the autocorrelation function, obtaining the autocorrelation data of the actual single-shot seismic data after multiple wave suppression; Step S110: replacing the autocorrelation data of the actual single-shot seismic data in step S105 and step S106 with the autocorrelation data of the actual single-shot seismic data after multiple wave suppression, replacing the actual single-shot seismic data in step S107 with the actual single-shot seismic data after multiple wave suppression, repeating steps S105 to S109 until the actual single-shot seismic data after multiple wave suppression is input into the trained support vector machine model in step S107, and only the actual single-shot seismic data without multiple waves is obtained; Step S111: Execute steps S105 to S110 for all actual single-shot seismic data on the common detection point gather of the actual seafloor seismograph data to obtain all actual single-shot seismic data without multiple waves.

2. The method according to claim 1, characterized in that The step S103 further comprises: before performing minimum phase wavelet shaping on the common detection point gathers of the actual seafloor seismograph data, preprocessing the actual seafloor seismograph data, wherein the preprocessing comprises bandpass filtering, random noise removal and normalization processing.

3. The method according to claim 1, characterized in that In step S105, the characteristic data includes peak position, peak amplitude, peak interval, decay speed and symmetry index.

4. The method according to claim 1, characterized in that: The step S105 comprises: Step S1051: Divide the characteristic data into a first training set and a first test set, wherein the ratio of characteristic data of autocorrelation data of simulated single shot seismic data to characteristic data of autocorrelation data of actual single shot seismic data in the first training set is 3:1, and the first test set is characteristic data of autocorrelation data of simulated single shot seismic data; Step S1052: Mark the first training set as a training set of feature data with multiple waves and feature data without multiple waves to train a support vector machine model, learn the boundary between the periodic characteristics and the non-periodic characteristics of the signal, and obtain a trained support vector machine model; use the first test set to evaluate the performance of the trained support vector machine model, and when the performance meets the standard, obtain a trained support vector machine model.

5. The method according to claim 1, characterized in that The step S106 includes: Step S1061: Divide the autocorrelation data of simulated single-shot seismic data and the autocorrelation data of actual single-shot seismic data into a second training set and a second test set, wherein the ratio of the autocorrelation data of simulated single-shot seismic data to the autocorrelation data of actual single-shot seismic data in the second training set is 3:1, and the second test set is the autocorrelation data of simulated single-shot seismic data; Step S1062: input the second training set into the input layer, use multiple convolutional layers with a step size of 2 to extract local features in the autocorrelation data, use a fully connected layer to integrate the local features and predict the delay time of the sub-peak, output the predicted delay time of the sub-peak in the output layer, and obtain the trained convolutional neural network model, wherein the predicted delay time of the sub-peak is a multiple wave period; use the second test set data to evaluate the performance of the convolutional neural network model, and when the performance meets the standard, obtain the trained convolutional neural network model.

6. A device for identifying and suppressing multiple waves of seafloor seismograph data, characterized in that: The device comprises: Forward simulation module: establish underground medium velocity model, set up OBS observation system, conduct earthquake forward simulation, and obtain simulated seafloor seismograph data; Data acquisition module: obtains actual seafloor seismograph data collected by field OBS stations; Minimum phase wavelet shaping module: performing minimum phase wavelet shaping on the common detection point gathers of the simulated seafloor seismograph data and the common detection point gathers of the actual seafloor seismograph data, respectively, to obtain the simulated seafloor seismograph data and the actual seafloor seismograph data that meet the minimum phase wavelet condition; The first autocorrelation module: using the autocorrelation function, autocorrelation processing is performed on the simulated seafloor seismograph data and the actual seafloor seismograph data that meet the minimum phase wavelet condition, to obtain the autocorrelation data set of the common detection point gather of the simulated seafloor seismograph data and the autocorrelation data set of the common detection point gather of the actual seafloor seismograph data; the autocorrelation data of the simulated single-shot seismic data and the autocorrelation data of the actual single-shot seismic data are obtained according to the autocorrelation data set of the common detection point gather of the simulated seafloor seismograph data and the autocorrelation data set of the common detection point gather of the actual seafloor seismograph data; Support vector machine model training module: extracting feature data of autocorrelation data of simulated single shot seismic data and autocorrelation data of actual single shot seismic data respectively, using the feature data to train a support vector machine model, and obtaining a trained support vector machine model; Convolutional neural network model training module: using the autocorrelation data of simulated single-shot seismic data and the autocorrelation data of actual single-shot seismic data to train the convolutional neural network model and obtain a trained convolutional neural network model; Multiple wave identification module: input the actual single shot seismic data into the trained support vector machine model to obtain the actual single shot seismic data with multiple waves and the actual single shot seismic data without multiple waves; Multiple wave suppression module: input the actual single shot seismic data with multiple waves into the trained convolutional neural network model to obtain the multiple wave periods of the actual single shot seismic data with multiple waves; use all the multiple wave periods of the actual single shot seismic data with multiple waves as prediction step lengths, test and determine prediction deconvolution parameters, suppress multiple waves by using prediction deconvolution according to the determined prediction deconvolution parameters, and obtain the actual single shot seismic data after the multiple waves are suppressed; The second autocorrelation module: using the autocorrelation function, obtains the autocorrelation data of the actual single-shot seismic data after multiple wave suppression; Loop module: replace the autocorrelation data of the actual single-shot seismic data in the support vector machine model training module and the convolutional neural network model training module with the autocorrelation data of the actual single-shot seismic data after multiple wave suppression, replace the actual single-shot seismic data in the multiple wave identification module with the actual single-shot seismic data after multiple wave suppression, and repeatedly run the support vector machine model training module, the convolutional neural network model training module, the multiple wave identification module, the multiple wave suppression module and the second autocorrelation module until the actual single-shot seismic data after multiple wave suppression is input into the trained support vector machine model in the multiple wave identification module, and only the actual single-shot seismic data without multiple waves is obtained; Repeated execution module: execute support vector machine model training module, convolutional neural network model training module, multiple wave identification module, multiple wave suppression module, second autocorrelation module and loop module for all single-shot seismic data on the common detection point gather of the actual seafloor seismometer data to obtain all actual single-shot seismic data without multiple waves.

7. An electronic device, characterized in that: include: processor; Memory; And a computer program, wherein the computer program is stored in the memory, and the computer program includes instructions, which, when executed by the processor, enable the electronic device to perform the steps of the method described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the method according to any one of claims 1 to 5.

Images