Diffraction multiple wave attenuation method, device, equipment and storage medium
By performing initial attenuation and neural network prediction on marine seismic data, effective attenuation of diffraction multiple waves is achieved, the problem of poor imaging quality in the prior art is solved, and the accuracy and efficiency of exploration results are improved.
Patent Information
- Application Number
- CN202210550247.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-05-20
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2042-05-20
AI Technical Summary
Existing filtering methods cannot effectively attenuate multiple diffraction waves in marine seismic data, resulting in poor imaging quality and affecting the accuracy of exploration results.
The primary attenuation is first performed to reduce the diffraction multiple wave data, and then the residual diffraction multiple wave existence area is predicted through neural network training, and the secondary attenuation is performed to obtain the primary wave data, which improves the accuracy and efficiency of data processing.
It effectively reduces interference from diffraction multiple waves, improves the imaging quality of seismic data, and ensures the accuracy and efficiency of exploration results.
Smart Images

Figure CN114895357B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of seismic data processing, and in particular to a diffraction multiple wave attenuation method, device, equipment, and storage medium. Background Art
[0002] In marine seismic exploration, since both the seabed and the sea surface are strong wave impedance interfaces, there is usually strong diffraction multiple wave interference in the seismic data. Since the existing filtering methods cannot effectively attenuate the diffraction multiple waves in the seismic data, the imaging quality of the seismic data is poor, affecting the accuracy of the exploration results. Summary of the invention
[0003] In view of the above problems, the present application provides a diffraction multiple wave attenuation method, device, equipment and storage medium, which can effectively attenuate the diffraction multiple waves in seismic data and ensure the accuracy of exploration results.
[0004] According to one aspect of the present application, a diffraction multiple wave attenuation method is provided, comprising: performing primary attenuation on diffraction multiple wave data in first original data to obtain first primary attenuation data; inputting training data into a neural network for training to obtain a prediction model for predicting a residual diffraction multiple wave existence region, the training data including the first primary attenuation data having a residual diffraction multiple wave region label; performing primary attenuation on diffraction multiple wave data in second original data to obtain second primary attenuation data; inputting detection data into the prediction model to obtain a residual diffraction multiple wave existence region in the detection data output by the prediction model, the detection data including the second primary attenuation data; performing secondary attenuation on the residual diffraction multiple wave data in the residual diffraction multiple wave existence region to obtain primary wave data.
[0005] The diffraction multiple wave attenuation method provided by the present application can preliminarily reduce the diffraction multiple waves in the first original data by firstly performing the primary attenuation on the diffraction multiple wave data in the first original data. Then, when the first primary attenuation data obtained by the primary attenuation is input into the neural network as training data for training, the computation amount of the neural network can be reduced and the computation speed can be improved. After training the training data, a prediction model for predicting the existence area of the residual diffraction multiple waves is obtained. After predicting the second primary attenuation data in the actual application by the prediction model, the existence area of the residual diffraction multiple waves in the second primary attenuation data can be accurately predicted first, so as to provide more accurate position conditions for the subsequent secondary attenuation of the residual diffraction multiple waves. Meanwhile, during the secondary attenuation, the residual diffraction multiple wave data only needs to be attenuated in the predicted existence area of the residual diffraction multiple waves, so as to effectively reduce the computation amount during the secondary attenuation and improve the efficiency of the secondary attenuation. Then, after performing the secondary attenuation on the residual diffraction multiple wave data in the predicted existence area of the residual diffraction multiple waves, the primary wave data can be accurately obtained. In the subsequent processing of the primary wave data, the accuracy of the exploration and analysis results can be effectively guaranteed.
[0006] In an optional manner, performing a primary attenuation on the diffraction multiple wave data in the first original data to obtain the first primary attenuation data includes: predicting the diffraction multiple wave data existing in the first original data to obtain the first predicted diffraction multiple wave data; subtracting the first predicted diffraction multiple wave data from the first original data to obtain the first primary attenuation data; and / or performing a primary attenuation on the diffraction multiple wave data in the second original data to obtain the second primary attenuation data includes predicting the diffraction multiple wave data existing in the second original data to obtain the second predicted diffraction multiple wave data; subtracting the second predicted diffraction multiple wave data from the second original data to obtain the second primary attenuation data. The present application firstly attenuates the first original data by using the SRME method, thereby preliminarily reducing the diffraction multiple wave data in the first original data, so that the amount of residual diffraction multiple wave data in the first initial attenuation data is reduced relative to that in the first original data, which is beneficial to reducing the amount of calculation of the computing device when predicting the residual diffraction multiple wave existence area through the neural network in the subsequent process, improving the prediction efficiency, and further performing secondary attenuation on the residual diffraction multiple wave data in the predicted residual diffraction multiple wave area, thereby achieving effective attenuation of the diffraction multiple waves in the detection data and improving the attenuation effect. The second original data is seismic wave data obtained during the actual operation of seismic petroleum exploration. The second original data is first attenuated by the SRME method to preliminarily reduce the diffraction multiple wave data in the second original data, so that the amount of calculation of the computing device can be reduced when the residual diffraction multiple wave existence area in the second original data is predicted by the neural network in the subsequent prediction, and the prediction efficiency is improved. In addition, the residual diffraction multiple wave data in the predicted residual diffraction multiple wave area is further attenuated twice to achieve effective attenuation of the diffraction multiple waves in the second original data, improve the quality of seismic wave imaging, and thus facilitate accurate analysis of stratigraphic structure, reservoir location, reservoir conditions, etc.
[0007] In an optional manner, after the first predicted diffraction multiple wave data is subtracted from the first original data to obtain the first initial attenuation data, the method further includes: performing reflected wave attenuation on the first predicted diffraction multiple wave data and / or the first initial attenuation data to obtain the first predicted diffraction multiple wave data with the reflected wave removed and / or the first initial attenuation data with the reflected wave removed; inputting the training data into a neural network for training to obtain a prediction model for predicting the area where the residual diffraction multiple waves exist, wherein the training data further includes the first predicted diffraction multiple wave data with the reflected wave removed and having a residual diffraction multiple wave area label and / or the first initial attenuation data with the reflected wave removed and having a residual diffraction multiple wave area label. Since the reflection wave is an important factor in analyzing the formation structure in the imaging data, in order to prevent the reflection wave from being mistakenly deleted due to the prediction model predicting the area where the reflection wave is located as the area where the residual multiple diffraction waves are located in the process of attenuating the multiple waves, the present application performs reflection wave attenuation on the first predicted diffraction multiple wave data and / or the first initial attenuation data, and inputs the obtained first predicted diffraction multiple wave data without the reflection wave and / or the first initial attenuation data without the reflection wave as training data into the neural network for training, not only when the amount of the first original data is constant, but also because the characteristics of the diffraction multiple waves in the first predicted diffraction multiple wave data without the reflection wave and / or the first initial attenuation data without the reflection wave are clearer, it is helpful to improve the training effect of the neural network and improve the accuracy of the prediction.
[0008] In an optional manner, after subtracting the second predicted diffraction multiple wave data from the second original data to obtain the second primary attenuation data, the method further includes: performing reflected wave attenuation on the second predicted diffraction multiple wave data and / or the second primary attenuation data to obtain the second predicted diffraction multiple wave data with the reflected wave removed and / or the second primary attenuation data with the reflected wave removed; inputting the detection data into the prediction model to obtain an area where residual diffraction multiple waves exist in the detection data output by the prediction model, wherein the detection data also includes the second predicted diffraction multiple wave data with the reflected wave removed and / or the second primary attenuation data with the reflected wave removed. Since the reflection wave is an important factor in analyzing the formation structure in the imaging data, in order to prevent the reflection wave from being mistakenly deleted due to the prediction model predicting the area where the reflection wave is located in the second initial attenuation data as the area where the residual multiple diffraction waves are located in the process of attenuating the multiple waves, thereby affecting the processing of subsequent data, the present application performs reflection wave attenuation on the second predicted diffraction multiple wave data and / or the second initial attenuation data, and inputs the obtained first predicted diffraction multiple wave data without the reflection wave and / or the first initial attenuation data without the reflection wave as prediction inputs into the neural network for prediction, thereby comprehensively and accurately analyzing the data received in the actual exploration process, that is, on the basis of the second original data, the second initial attenuation data obtained according to the second original data is analyzed. , the second predicted diffraction multiple wave data after the reflection wave is removed and / or the second initial attenuation data after the reflection wave is removed all predict the area where the residual diffraction multiple waves exist, which is beneficial to analyzing the accuracy of the final prediction result. For example, when the residual diffraction multiple wave existence area predicted in the second initial attenuation data and the residual diffraction multiple wave existence area predicted in the second initial attenuation data after the reflection wave are highly overlapped, it means that the prediction result is relatively accurate. When the residual diffraction multiple wave existence area predicted in the second initial attenuation data and the residual diffraction multiple wave existence area predicted in the second initial attenuation data after the reflection wave are greatly different and the overlap is poor, the prediction result is not accurate enough and there may be other interference factors, which may require human intervention for analysis.
[0009] In an optional manner, the residual diffraction multiple wave data in the residual diffraction multiple wave existence area is subjected to secondary attenuation to obtain the primary wave data, including: obtaining the offset distance corresponding to the detection data, the offset distance being the distance between the signal transmitting device and the signal receiving device for obtaining the second original data; determining the window width according to the offset distance; and performing secondary attenuation on the residual diffraction multiple wave data according to the determined window width. Since the corresponding maximum offset distance and minimum offset distance in multiple detection data may vary in different exploration operations, by setting the first window width and the second window width according to the maximum offset distance and the minimum offset distance, it is ensured that when performing secondary attenuation on different detection data, different first window widths and second window widths can be set according to the specific conditions of the maximum offset distance and the minimum offset distance in different detection data, so as to realize the reasonable matching and use of the window width during the subsequent secondary attenuation. And by using the linear interpolation method to determine the window width matching other detection data according to the maximum offset distance, the minimum offset distance, the first window width and the second window width, the adaptive matching of the window width is realized, so that the use of the window width is more reasonable, and the amount of data input manually on the computing device can be effectively reduced, thereby improving efficiency.
[0010] In an optional manner, the residual diffraction multiple wave data in the residual diffraction multiple wave existing area is subjected to secondary attenuation to obtain primary wave data, including: learning the diffraction multiple wave features in the first predicted diffraction multiple wave data; using the learned diffraction multiple wave features to establish a diffraction multiple wave model according to the residual diffraction multiple wave data; subtracting the diffraction multiple wave model from the residual diffraction multiple wave data to obtain the primary wave data. After secondary attenuation using the dictionary feature learning method, the residual diffraction multiple waves in the image are significantly reduced, and the primary wave data are clearer, thereby facilitating subsequent analysis and processing.
[0011] According to another aspect of the present application, a diffraction multiple wave attenuation device is provided, comprising: a first primary attenuation module, used for performing primary attenuation on diffraction multiple wave data in first original data to obtain first primary attenuation data; a training module, used for inputting training data into a neural network for training to obtain a prediction model for predicting a residual diffraction multiple wave existence region, wherein the training data includes the first primary attenuation data having a residual diffraction multiple wave region label; a second primary attenuation module, used for performing primary attenuation on diffraction multiple wave data in second original data to obtain second primary attenuation data; a prediction module, used for inputting detection data into the prediction model to obtain a residual diffraction multiple wave existence region in the detection data output by the prediction model, wherein the detection data includes the second primary attenuation data; and a secondary attenuation module, used for performing secondary attenuation on the residual diffraction multiple wave data in the residual diffraction multiple wave existence region to obtain primary wave data.
[0012] According to another aspect of the present application, a computing device is provided, comprising: a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; the memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute the diffraction multiple wave attenuation method as described in any of the above items.
[0013] According to another aspect of the present application, a computer storage medium is provided, wherein at least one executable instruction is stored in the storage medium, and the executable instruction enables a processor to execute the diffraction multiple wave attenuation method as described in any one of the above items.
[0014] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present application. Also, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:
[0016] Figure 1 A flow chart of a diffraction multiple wave attenuation method provided in an embodiment of the present application;
[0017] Figure 2a This is a schematic diagram of an image of the second initial attenuation data in the input prediction model provided in an embodiment of the present application;
[0018] Figure 2b An image of the second primary attenuation data with the area where the residual diffraction multiple waves exist, output by the prediction model provided in an embodiment of the present application;
[0019] Figure 3 A flow chart of a diffraction multiple wave attenuation method provided by another embodiment of the present application;
[0020] Figure 4 for Figure 1 Flow chart of sub-steps of step S150;
[0021] Figure 5 for Figure 4 Flow chart of sub-steps of step S152;
[0022] Figure 6a The data image after secondary attenuation using a relatively large fixed window width provided in the embodiment of the present application;
[0023] Figure 6b A data image after secondary attenuation using a smaller fixed window width provided in an embodiment of the present application;
[0024] Figure 6c A data image after secondary attenuation using an adaptive window width provided in an embodiment of the present application;
[0025] Figure 7 for Figure 1 Another sub-step flow chart of step S150;
[0026] Figure 8a An image before secondary attenuation of residual diffraction multiple wave data in an area where residual diffraction multiple waves exist provided in an embodiment of the present application;
[0027] Figure 8b The dictionary feature learning method provided in the embodiment of the present application is used to Figure 8a The image after the image in is subjected to secondary attenuation;
[0028] Fig. 9 A schematic diagram of the structure of a diffraction multiple wave attenuation device provided in an embodiment of the present application;
[0029] Fig.10 A schematic diagram of the structure of a computing device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0030] Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein.
[0031] In marine seismic exploration, the operating ship first arranges underwater equipment such as collection tow cables and water birds on the seabed, and then uses compressed air guns or explosives as seismic sources to excite seismic waves. The seismic waves pass through the seawater and enter the formation. The detectors on the collection tow cables receive the seismic data reflected back from each rock layer, and then the reflected seismic data are processed to determine the location of the oil reservoir, reservoir conditions, etc.
[0032] The imaging quality of seismic data directly affects the accuracy of exploration results. Since both the seabed and the sea surface are strong impedance interfaces, there is usually strong diffraction multiple wave interference in the seismic data, and diffraction multiple waves are an important factor affecting the imaging quality of seismic data.
[0033] As oil and gas exploration continues to move towards deepwater environments, the rugged seabed environment increases and the diffraction multiple waves become more and more complicated. As a result, the diffraction multiple waves cannot be effectively attenuated, which seriously restricts the imaging quality of seismic data.
[0034] Based on this, the present application proposes a diffraction multiple wave attenuation method, by first performing a primary attenuation on the diffraction multiple wave data in the first original data, the diffraction multiple waves in the first original data can be preliminarily reduced, and then when the first primary attenuation data obtained by the primary attenuation is input into the neural network as training data for training, the amount of calculation of the neural network can be reduced and the calculation speed can be improved. After training the training data, a prediction model for predicting the existence area of residual diffraction multiple waves is obtained. After predicting the second primary attenuation data in the actual application by the prediction model, the existence area of residual diffraction multiple waves in the second primary attenuation data can be accurately predicted first, and a more accurate position condition is provided for the subsequent secondary attenuation of residual diffraction multiple waves. At the same time, during the secondary attenuation, the residual diffraction multiple wave data only needs to be attenuated in the predicted residual diffraction multiple wave existence area, which can effectively reduce the amount of calculation during the secondary attenuation and improve the efficiency of the secondary attenuation. After the residual diffraction multiple wave data in the predicted residual diffraction multiple wave existence area is secondary attenuated, the primary wave data can be accurately obtained, and the subsequent processing of the primary wave data can effectively ensure the accuracy of the exploration and analysis results.
[0035] See also Figure 1 , the figure shows the process of the diffraction multiple wave attenuation method provided by an embodiment of the present application, the method is executed by a computing device that needs to perform diffraction multiple wave attenuation, such as a mobile phone, a computer, a server, etc. As shown in the figure, the method includes:
[0036] S110: Performing initial attenuation on the diffraction multiple wave data in the first original data to obtain first initial attenuation data.
[0037] In this step, the first original data is the seismic data received by the signal receiving device (such as a detector) on the acquisition tow cable. The signal transmitting device (i.e., the seismic source, such as a compressed air gun or explosives) excites seismic waves, which are reflected by the water surface and the stratum. The signal receives the reflected seismic waves to form seismic data.
[0038] Specifically, the diffraction multiple waves in the first original data can be initially attenuated by utilizing the difference between the primary wave and the multiple waves in the transform domain. For example, the attenuation of the diffraction multiple waves can be achieved by using FK filtering (frequency wavenumber filtering), Radon transform, velocity-based filtering, predictive subtraction, etc.
[0039] S120: Inputting the training data into the neural network for training to obtain a prediction model for predicting the area where the residual diffraction multiple waves exist, the training data including the first initial attenuation data having the residual diffraction multiple wave area label.
[0040] The neural network uses a target detection network to identify specific targets in the image. The target detection network will find the trained target in the image and output the target's bounding box as a result. Specifically, the number of the first initial attenuation data is multiple. First, the area where the residual diffraction multiple waves exist is marked on each first initial attenuation data as a label for neural network training, and then the first initial attenuation data is input into the neural network for training.
[0041] The loss function used by the neural network is Loss = L obj +L box +L clc , where L clc Represents the classification loss function. Since the neural network used only needs to classify the residual diffraction multiple wave, L clc =0, then Loss = L obj +L box .
[0042] L obj Represents the confidence loss function, which ensures that the neural network's predictions of location information and category information are not completely independent. Its expression is:
[0043]
[0044] Among them, λ noobj and λ obj are the proportionality coefficients of the two terms. 2 Represents the number of grids the entire image is divided into. B represents all detected areas. It means that if there is a target in the jth detected area in the i-th grid, the value is 0, and if there is no target, the value is 1. Indicates that if there is a target in the jth detected area in the i-th grid, the value is 1, and if there is no target, the value is 0. i Represents the true confidence probability, which is 1 if there is a target in the grid, otherwise it is 0. is the confidence probability obtained based on the network prediction results.
[0045] L box It is the border loss function, which represents the difference between the predicted target box position and the target box position given in the annotation, and is used to constrain the accuracy of position detection. Its expression is:
[0046]
[0047] Among them, λ coord is the proportionality coefficient. 2 Represents the number of grids the entire image is divided into. B represents all detected areas. Indicates that if there is a target in the jth detected area in the i-th grid, the value is 1, and if there is no target, the value is 0. i ,y i 、w i 、h i They represent the horizontal coordinate of the border center point, the vertical coordinate of the center point, the border width, and the border height given in the label respectively. They represent the horizontal coordinate of the center point of the border, the vertical coordinate of the center point, the width of the border, and the height of the border respectively.
[0048] S130: Performing primary attenuation on the diffraction multiple wave data in the second original data to obtain second primary attenuation data.
[0049] In this step, the second original data is seismic data received by the signal receiving device on the streamer during the actual oil exploration operation. Specifically, the attenuation of the diffracted multiple waves in the second original data can be achieved by FK filtering, Radon transform, velocity-based filtering, predictive subtraction, etc.
[0050] S140: Input the detection data into the prediction model to obtain the region where the residual diffraction multiple waves exist in the detection data output by the prediction model, wherein the detection data includes the second initial attenuation data.
[0051] In this step, after the second primary attenuation data is input into the trained prediction model, the prediction model will output the area where the residual diffraction multiple waves in the second primary attenuation data exist, providing conditions for subsequent secondary attenuation of the residual diffraction multiple waves.
[0052] Please refer to Figure 2a and Figure 2b , Figure 2a The second primary attenuation data used to input the prediction model. The area in the frame in the figure represents the area containing residual diffraction multiple waves in the second primary attenuation data. It should be noted that Figure 2a The box in is marked in the figure for the convenience of explaining the position of the residual diffraction multiple waves. This box does not exist in the second primary attenuation data actually input into the prediction model. Figure 2b The data is output by the prediction model, which is the second initial attenuation data with the area where the residual diffraction multiple waves are predicted to exist by the prediction model.
[0053] S150: performing secondary attenuation on the residual diffraction multiple wave data in the region where the diffraction multiple waves exist, to obtain primary wave data.
[0054] In this step, by performing secondary attenuation on the residual diffraction multiple wave data in the region where the residual diffraction multiple waves in the second initial attenuation data determined in step S140 exist, the diffraction multiple waves in the second initial data are further reduced, so that the final primary wave data is more accurate, thereby facilitating the subsequent analysis of the formation conditions. Specifically, the secondary attenuation can also adopt the above-mentioned FK filtering method, Radon transform method, velocity-based filtering method, predictive subtraction method, etc., to achieve further attenuation of the diffraction multiple waves in the second initial data.
[0055] The diffraction multiple wave attenuation method provided by the present application can preliminarily reduce the diffraction multiple waves in the first original data by firstly performing the primary attenuation on the diffraction multiple wave data in the first original data. Then, when the first primary attenuation data obtained by the primary attenuation is input into the neural network as training data for training, the computation amount of the neural network can be reduced and the computation speed can be improved. After training the training data, a prediction model for predicting the existence area of the residual diffraction multiple waves is obtained. After predicting the second primary attenuation data in the actual application by the prediction model, the existence area of the residual diffraction multiple waves in the second primary attenuation data can be accurately predicted first, so as to provide more accurate position conditions for the subsequent secondary attenuation of the residual diffraction multiple waves. Meanwhile, during the secondary attenuation, the residual diffraction multiple wave data only needs to be attenuated in the predicted existence area of the residual diffraction multiple waves, so as to effectively reduce the computation amount during the secondary attenuation and improve the efficiency of the secondary attenuation. Then, after performing the secondary attenuation on the residual diffraction multiple wave data in the predicted existence area of the residual diffraction multiple waves, the primary wave data can be accurately obtained. In the subsequent processing of the primary wave data, the accuracy of the exploration and analysis results can be effectively guaranteed.
[0056] For the above step S110, this application further proposes an implementation method, please refer to Figure 3 , the figure shows the process of the diffraction multiple wave attenuation method provided by another embodiment. As shown in the figure, step S110 includes:
[0057] S111: predicting diffraction multiple wave data present in the first original data to obtain first predicted diffraction multiple wave data;
[0058] S112: Subtract the first diffraction multiple wave data from the first original data to obtain first initial attenuation data.
[0059] In this scheme, the diffraction multiple waves are initially attenuated by the prediction subtraction method, specifically, the diffraction multiple wave model (i.e., the first predicted diffraction multiple wave data) is first predicted from the first original data, and then the predicted diffraction multiple wave model is adaptively subtracted from the first original data, thereby achieving the attenuation of the diffraction multiple waves.
[0060] The prediction subtraction method can use the free surface multiple wave attenuation method (surface related multiple elimination, referred to as SRME). The SRME method is divided into two steps. The first step is to predict the existing diffraction multiple waves from the seismic data. The second step is to adaptively subtract the predicted diffraction multiple waves from the seismic data. In the second step, the matched filter method is usually used to subtract the diffraction multiple waves, that is, to solve min||D-FM|| L Such an optimization problem, where D is the original data, M is the diffraction multiple waves predicted in the first step of the SRME method, F is the required matched filter parameter, and L is the norm of the optimization problem, generally L = 1 or L = 2. Since the subtraction effect of the matched filter depends largely on the accuracy of the diffraction multiple waves predicted in the first step, and there may be large differences in amplitude and phase between the predicted diffraction multiple waves and the actual diffraction multiple waves, the effect of attenuating the diffraction multiple waves only by the SRME method is poor.
[0061] Based on the above problems, the present application first performs a primary attenuation on the first original data through the SRME method, thereby preliminarily reducing the diffraction multiple wave data in the first original data, so that the amount of residual diffraction multiple wave data in the first primary attenuation data is reduced relative to that in the first original data, which is beneficial to reducing the amount of calculation of the computing device when predicting the residual diffraction multiple wave existence area through the neural network in the subsequent process, thereby improving the prediction efficiency, and further performing a secondary attenuation on the residual diffraction multiple wave data in the predicted residual diffraction multiple wave area, thereby achieving effective attenuation of the diffraction multiple waves in the detection data and improving the attenuation effect.
[0062] For the above step 130, this application further proposes an implementation method, please refer to Figure 3 As shown in the figure, step S130 includes:
[0063] S131: predicting the diffraction multiple wave data present in the second original data to obtain second predicted diffraction multiple wave data;
[0064] S132: Subtract the second diffraction multiple wave data from the second original data to obtain second primary attenuation data.
[0065] Similar to the method of performing initial attenuation on the first original data in the above step S110, the diffraction multiple waves in the second original data are also attenuated by the prediction subtraction method in the present solution. Specifically, a diffraction multiple wave model (i.e., second predicted diffraction multiple wave data) is first predicted from the second original data, and then the predicted diffraction multiple wave model is adaptively subtracted from the second original data, thereby achieving attenuation of the diffraction multiple waves.
[0066] The second original data is seismic wave data obtained during the actual operation of seismic petroleum exploration. The second original data is first attenuated by the SRME method to preliminarily reduce the diffraction multiple wave data in the second original data, so that the amount of calculation of the computing device can be reduced when the residual diffraction multiple wave existence area in the second original data is predicted by the neural network in the subsequent prediction, and the prediction efficiency is improved. In addition, the residual diffraction multiple wave data in the predicted residual diffraction multiple wave area is further attenuated twice to achieve effective attenuation of the diffraction multiple waves in the second original data, improve the quality of seismic wave imaging, and thus facilitate accurate analysis of stratigraphic structure, reservoir location, reservoir conditions, etc.
[0067] In order to improve the accuracy of neural network prediction, this application further proposes an implementation method, please refer to Figure 3 As shown in the figure, after the above step S112, the method further includes:
[0068] S113: performing reflected wave attenuation on the first predicted diffraction multiple wave data and / or the first initial attenuation data to obtain the first predicted diffraction multiple wave data with reflected waves removed and / or the first initial attenuation data with reflected waves removed.
[0069] In this step, the reflected wave can be attenuated by utilizing the difference in angle and shape between the reflected wave and other waves.
[0070] In sub-step S121 of the above step S120, the training data further includes first predicted diffraction multiple wave data with residual diffraction multiple wave region labels after removal of reflected waves and / or first initial attenuation data with residual diffraction multiple wave region labels after removal of reflected waves.
[0071] Since the reflection wave is an important factor in analyzing the formation structure in the imaging data, in order to prevent the reflection wave from being mistakenly deleted due to the prediction model predicting the area where the reflection wave is located as the area where the residual multiple diffraction waves are located in the process of attenuating the multiple waves, the present application performs reflection wave attenuation on the first predicted diffraction multiple wave data and / or the first initial attenuation data, and inputs the obtained first predicted diffraction multiple wave data without the reflection wave and / or the first initial attenuation data without the reflection wave as training data into the neural network for training, not only when the amount of the first original data is constant, but also because the characteristics of the diffraction multiple waves in the first predicted diffraction multiple wave data without the reflection wave and / or the first initial attenuation data without the reflection wave are clearer, it is helpful to improve the training effect of the neural network and improve the accuracy of the prediction.
[0072] In order to comprehensively and accurately analyze the data received during the actual exploration process, this application further proposes an implementation method. Please refer to Figure 3As shown in the figure, after the above step S132, the method further includes:
[0073] S133: performing reflected wave attenuation on the second predicted diffraction multiple wave data and / or the second initial attenuation data to obtain the second predicted diffraction multiple wave data with reflected waves removed and / or the second initial attenuation data with reflected waves removed.
[0074] Likewise, in this step, the reflected wave can be attenuated by utilizing the angle and shape differences between the reflected wave and other waves.
[0075] In the sub-step S141 of the above step S140, the detection data further includes the second predicted diffraction multiple wave data after removing the reflected wave and / or the second primary attenuation data after removing the reflected wave.
[0076] Since the reflection wave is an important factor in analyzing the formation structure in the imaging data, in order to prevent the reflection wave from being mistakenly deleted due to the prediction model predicting the area where the reflection wave is located in the second initial attenuation data as the area where the residual multiple diffraction waves are located in the process of attenuating the multiple waves, thereby affecting the processing of subsequent data, the present application performs reflection wave attenuation on the second predicted diffraction multiple wave data and / or the second initial attenuation data, and inputs the obtained first predicted diffraction multiple wave data without the reflection wave and / or the first initial attenuation data without the reflection wave as prediction inputs into the neural network for prediction, thereby comprehensively and accurately analyzing the data received in the actual exploration process, that is, on the basis of the second original data, the second initial attenuation data obtained according to the second original data is analyzed. , the second predicted diffraction multiple wave data after the reflection wave is removed and / or the second initial attenuation data after the reflection wave is removed all predict the area where the residual diffraction multiple waves exist, which is beneficial to analyzing the accuracy of the final prediction result. For example, when the residual diffraction multiple wave existence area predicted in the second initial attenuation data and the residual diffraction multiple wave existence area predicted in the second initial attenuation data after the reflection wave are highly overlapped, it means that the prediction result is relatively accurate. When the residual diffraction multiple wave existence area predicted in the second initial attenuation data and the residual diffraction multiple wave existence area predicted in the second initial attenuation data after the reflection wave are greatly different and the overlap is poor, the prediction result is not accurate enough and there may be other interference factors, which may require human intervention for analysis.
[0077] For the above step S150, this application further proposes an implementation method, please refer to Figure 4 , the figure shows the process of step S150 sub-steps. As shown in the figure, step S150 includes:
[0078] S151: Acquire an offset distance corresponding to the detection data, where the offset distance is the distance between the signal transmitting device and the signal receiving device that acquires the second original data.
[0079] In this step, the signal transmitting device is a device such as explosives, air guns, etc. for exciting seismic waves, and the signal receiving device is a detector arranged on the acquisition tow cable. The offset distance is the distance between the explosives or air guns that excite seismic waves and the detector that receives the second original data. In petroleum seismic exploration operations, the offset distance can be determined in advance, and the technicians arrange the signal transmitting device and the signal receiving device accordingly according to the determined offset distance. Of course, the offset distance can also be measured by the measuring equipment after the arrangement. The offset distance determined or measured by the measuring equipment is input into the calculation device to achieve the acquisition of the offset distance.
[0080] The energy of the diffracted multiple waves is different at different offset distances. When the offset distance between the signal receiving device and the signal transmitting device is small, the energy of the diffracted multiple waves is stronger, and vice versa.
[0081] S152: Determine the window width according to the offset;
[0082] S153: Perform secondary attenuation on the residual diffraction multiple wave data according to the determined window width.
[0083] The window width indicates the size (length, width, radius, diameter, etc.) or area of the window for attenuating the residual diffraction multiple wave data when performing secondary attenuation in the data predicted by the prediction model. The size of the window width can control the attenuation degree of the residual diffraction multiple wave data. When the window width is large, the attenuation rate is fast, but the attenuation degree is light. On the contrary, when the window width is small, the attenuation rate is slow and the attenuation degree is heavy. Generally, the smaller the offset distance, the stronger the energy of the diffraction multiple wave, and the smaller the window width can be set to ensure that the diffraction multiple wave is attenuated to a greater extent. The larger the offset distance is, the larger the window width can be set, so as to effectively attenuate the diffraction multiple wave with weaker energy while increasing the attenuation rate.
[0084] In steps S152 and S153, the window widths corresponding to different offset ranges can be preset in advance. Then, after obtaining the offset corresponding to the residual diffracted multiple data, according to the range where the value of the offset lies, the corresponding window width is automatically matched for the subsequent secondary attenuation processing. For example, it can be set that when the offset d ≤ 200m, the window width is 30 (i.e., the window for secondary attenuation is a square of 30 samples × 30 samples or 30 traces × 30 traces), when 200 < d ≤ 400m, the window width is 40, when 400 < d ≤ 600m, the window width is 50, when 600 < d ≤ 800m, the window width is 60, and when 800 < d ≤ 1000m, the window width is 70. The above is only a specific window width setting method provided for the convenience of explanation and is not used to limit the relationship between the window width and the offset. In other embodiments, the setting of the window width can be adjusted accordingly according to the actual operation situation, which will not be elaborated here.
[0085] Determine the window width according to the offset, and perform Russian secondary attenuation on the residual diffracted multiple data according to the determined window width to achieve the attenuation processing with an adaptive window width. Thus, when the energy of the diffracted multiple data is strong, the effective attenuation of the diffracted multiple data can be ensured. When the energy of the diffracted multiple data is weak, while ensuring the effective attenuation of the diffracted multiple data, the attenuation rate can be accelerated.
[0086] For the above step S152, the present application further proposes an implementation manner. For details, please refer to Figure 5 , which shows the flow of the sub-steps of step S152. The number of detection data is multiple, and the multiple predicted wave data correspond to at least two different sizes of offsets. Step S152 includes:
[0087] S1521: Obtain the first window width corresponding to the maximum offset and the second window width corresponding to the minimum offset.
[0088] Since in the process of petroleum seismic exploration, multiple geophones are arranged on the acquisition streamer, each geophone receives the second original data, and the offset corresponding to each geophone is also different. That is, the second original data received by different geophones, and the multiple detection data processed from the second original data, their corresponding offsets are also different. Therefore, in this step, the maximum offset and the minimum offset respectively refer to the maximum value and the minimum value of the multiple offsets corresponding to the multiple detection data input to the prediction model.
[0089] S1522: Use the linear interpolation method to determine the window width corresponding to the offset between the maximum offset and the minimum offset.
[0090] In actual operations, the acquisition towline is usually several kilometers or even more than ten kilometers long, and a large number of detectors are arranged on the acquisition towline. In order to facilitate the rapid setting of the window width and improve the matching degree of the window frame under different offset distances, this scheme uses linear interpolation to determine the window width corresponding to the offset distance between the maximum offset distance and the minimum offset distance.
[0091] Specifically, after the computing device obtains the maximum offset and the minimum offset in a plurality of detection data, the technician specifically inputs the first window width matching the maximum offset and the second window width matching the minimum offset in the computing device according to the specific values of the maximum offset and the minimum offset. In addition to the detection data corresponding to the maximum offset and the detection data corresponding to the minimum offset, the window width for secondary attenuation of other detection data is determined by the computing device using linear interpolation according to the maximum offset, the minimum offset, the first window width and the second window width. Generally, in order to ensure the reliability of secondary attenuation, the first window width is the maximum window width and the second window width is the minimum window width.
[0092] Since the corresponding maximum offset and minimum offset in multiple detection data may vary in different exploration operations, the first window width and the second window width are set according to the maximum offset and the minimum offset, which ensures that when performing secondary attenuation on different detection data, different first window widths and second window widths can be set according to the specific conditions of the maximum offset and the minimum offset in different detection data, so as to achieve reasonable matching and use of the window width during subsequent secondary attenuation. In addition, by using the linear interpolation method to determine the window width matching other detection data according to the maximum offset, the minimum offset, the first window width and the second window width, adaptive matching of the window width is achieved, making the use of the window width more reasonable, and can effectively reduce the amount of data input manually on the computing device, thereby improving efficiency.
[0093] See also Figure 6a-6c For the convenience of explanation, Figure 6a The results of using a fixed larger window width for attenuation are shown in Figure 6b The processing result of using a fixed small window width for attenuation is shown in Figure 6c The processing result of attenuation using adaptive window width in the present application is shown in FIG. As can be seen from the figure, when the residual diffraction multiple wave energy is strong, the attenuation effect of a fixed larger window width is relatively poor, while the attenuation effect of an adaptive window width is substantially the same as that of a fixed smaller window width. When the residual diffraction multiple wave energy is weak and the area of the attenuation region is certain, the fixed minimum window width has many attenuation times and a large amount of calculation, which will result in a slow attenuation speed. When the residual diffraction multiple wave energy is weak, the adaptive window width used in the present application has a larger window width correspondingly and the area of the attenuation region is certain, which can simultaneously ensure the attenuation effect and attenuation speed.
[0094] For the above step S150, this application further proposes an implementation method, please refer to Figure 7 , another process of step S150 sub-step is shown in the figure. As shown in the figure, step S150 includes:
[0095] S154: Learning the diffraction multiple wave characteristics in the first predicted diffraction multiple wave data.
[0096] In this step, the dictionary feature learning method is used to learn the key features (i.e., diffraction multiple wave features) in the first predicted diffraction multiple wave data. The multiple wave codebook B learned from the first predicted diffraction multiple wave data is:
[0097]
[0098] subject to B T B=I
[0099] Where M is the first predicted diffraction multiple wave data, and X is the eigenvector.
[0100] S155: Using the learned diffraction multiple wave characteristics, a diffraction multiple wave model is established according to the residual diffraction multiple wave data.
[0101] In this step, the diffraction multiple wave characteristics learned in step S154 are used to reconstruct the diffraction multiple wave model according to the residual diffraction multiple wave data in the residual diffraction wave existing area in the detection data. The reconstructed diffraction multiple wave model Y is:
[0102]
[0103] M'=BY
[0104] Where S is the data containing residual diffraction multiple waves.
[0105] S156: Subtract the diffraction multiple wave model from the residual diffraction multiple wave data to obtain the primary wave data.
[0106] In this step, the primary wave data P obtained is:
[0107] P=SM′
[0108] See also Figure 8a and Figure 8b , in order to facilitate the description of this scheme, Figure 8a and Figure 8b, respectively, show the image before secondary attenuation of the residual diffraction multiple wave data in the area where the residual diffraction multiple wave exists and the image after secondary attenuation of the residual diffraction multiple wave data in the area where the residual diffraction multiple wave exists using the dictionary feature learning method. It can be seen from the figure that after secondary attenuation using the dictionary feature learning method, the residual diffraction multiple waves in the image are significantly reduced, and the primary wave data are clearer, which is convenient for subsequent analysis and processing.
[0109] It should be noted that when performing secondary attenuation on the residual diffraction multiple wave data in the region where the residual diffraction multiple waves exist, in some embodiments, the following method is used: Figure 4 The window width of the secondary attenuation is determined by the method of steps S151 to S153 in the above, and the secondary attenuation method can adopt FK filtering method, Radon transform method, velocity-based filtering method, prediction subtraction method, etc., and the diffraction multiple wave data can be secondary attenuated in the determined window width by any of the above methods. In other embodiments, the following method can be directly used in the predicted residual diffraction multiple wave existence area: Figure 7 The graph dictionary feature learning method shown in step S154 to step S156 in the above embodiment performs secondary attenuation on the diffraction multiple wave data. In some other embodiments, in the following example, Figure 4 Based on the method of steps S151 to S153 in the above, the window width of the secondary attenuation is determined as follows: Figure 7 The graph dictionary feature learning method shown in steps S154 to S156 performs secondary attenuation on the diffraction multiple wave data, and in this method, there is no sequential relationship between steps S151, S152 and steps S154, S155, and steps S153 and S156 are combined, specifically, according to the determined window width, the diffraction multiple wave model is subtracted from the residual diffraction multiple wave data in the window width to obtain the primary wave data. The primary wave data obtained by this method is clearer and more accurate, which is conducive to subsequent analysis and processing.
[0110] According to another aspect of the present invention, a diffraction multiple wave attenuation device is provided. Fig. 9, the figure shows the structure of the diffraction multiple wave attenuation device provided by an embodiment. As shown in the figure, the diffraction multiple wave attenuation device 200 includes a first primary attenuation module 210, a training module 220, a second primary attenuation module 230, a prediction module 240, and a secondary attenuation module 250. The first primary attenuation module 210 is used to perform primary attenuation on the diffraction multiple wave data in the first original data to obtain the first primary attenuation data. The training module 220 is used to input the training data into the neural network for training to obtain a prediction model for predicting the residual diffraction multiple wave existence area, and the training data includes the first primary attenuation data with the residual diffraction multiple wave area label. The second primary attenuation module 230 is used to perform primary attenuation on the diffraction multiple wave data in the second original data to obtain the second primary attenuation data. The prediction module 240 is used to input the detection data into the prediction model to obtain the residual diffraction multiple wave existence area in the detection data output by the prediction model, and the detection data includes the second primary attenuation data. The secondary attenuation module 250 is used to perform secondary attenuation on the residual diffraction multiple wave data in the diffraction multiple wave existence area to obtain the primary wave data.
[0111] In some embodiments, the first initial attenuation module 210 is used to predict the diffraction multiple wave data present in the first original data to obtain first predicted diffraction multiple wave data, and to subtract the first diffraction multiple wave data from the first original data to obtain first initial attenuation data.
[0112] In some embodiments, the second primary attenuation module 230 is used to predict the diffraction multiple wave data present in the second original data to obtain second predicted diffraction multiple wave data, and is used to subtract the second diffraction multiple wave data from the second original data to obtain second primary attenuation data.
[0113] Please continue reading Fig. 9 In some embodiments, the diffraction multiple wave attenuation device 200 further includes a first reflection wave attenuation module 260, which is used to perform reflection wave attenuation on the first predicted diffraction multiple wave data and / or the first initial attenuation data to obtain the first predicted diffraction multiple wave data with the reflection wave removed and / or the first initial attenuation data with the reflection wave removed. The training module 220 is used to input the training data into the neural network for training to obtain a prediction model for predicting the residual diffraction multiple wave existence area, and the training data includes the first initial attenuation data with the residual diffraction multiple wave area label, the first predicted diffraction multiple wave data with the reflection wave removed with the residual diffraction multiple wave area label, and / or the first initial attenuation data with the reflection wave removed with the residual diffraction multiple wave area label.
[0114] Please continue reading Fig. 9In some embodiments, the diffraction multiple wave attenuation device 200 further includes a second reflected wave attenuation module 270, which is used to perform reflected wave attenuation on the second predicted diffraction multiple wave data and / or the second initial attenuation data to obtain the second predicted diffraction multiple wave data with the reflected wave removed and / or the second initial attenuation data with the reflected wave removed. The prediction module 240 is used to input the detection data into the prediction model to obtain the residual diffraction multiple wave existence area in the detection data output by the prediction model, and the detection data includes the second initial attenuation data, the second predicted diffraction multiple wave data with the reflected wave removed and / or the second initial attenuation data with the reflected wave removed.
[0115] In some embodiments, the secondary attenuation module 250 is used to obtain an offset distance corresponding to the detection data, where the offset distance is the distance between a signal transmitting device and a signal receiving device that obtains the second original data, and is used to determine a window width according to the offset distance, and to perform secondary attenuation on the residual diffraction multiple wave data according to the determined window width.
[0116] In some embodiments, the number of detection data is multiple, and the multiple predicted wave data correspond to at least two offsets of different sizes. The secondary attenuation module 250 is used to obtain a first window width corresponding to the maximum offset and a second window width corresponding to the minimum offset, and to determine the window width corresponding to the offset between the maximum offset and the minimum offset by using a linear interpolation method.
[0117] In some embodiments, the secondary attenuation module 250 is used to learn the diffraction multiple wave characteristics in the first predicted diffraction multiple wave data, to use the learned diffraction multiple wave characteristics to establish a diffraction multiple wave model according to the residual diffraction multiple wave data, and to subtract the diffraction multiple wave model from the residual diffraction multiple wave data to obtain the primary wave data.
[0118] According to another aspect of the embodiment of the present application, a computing device is also provided. Fig.10 , the figure shows the structure of a computing device provided by an embodiment, and the specific embodiments of the present application do not limit the specific implementation of the computing device.
[0119] like Fig.10 As shown, the computing device may include: a processor (processor) 302 , a communication interface (Communications Interface) 303 , a memory (memory) 306 , and a communication bus 308 .
[0120] The processor 302, the communication interface 303, and the memory 306 communicate with each other via the communication bus 308. The communication interface 303 is used to communicate with other devices such as a client or other server network elements. The processor 302 is used to execute the program 310, which can specifically execute the relevant steps in the above-mentioned embodiment of the method for attenuating diffraction multiple waves.
[0121] Specifically, the program 310 may include program code including computer executable instructions.
[0122] The processor 302 may be a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present invention. The one or more processors included in the computing device may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.
[0123] The memory 306 is used to store the program 310. The memory 306 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.
[0124] The program 310 may be specifically invoked by the processor 302 to cause the computing device to perform the following operations:
[0125] Performing a primary attenuation on the diffraction multiple wave data in the first original data to obtain first primary attenuation data;
[0126] Inputting training data into a neural network for training to obtain a prediction model for predicting a region where residual diffraction multiple waves exist, the training data including first primary attenuation data having a residual diffraction multiple wave region label;
[0127] Performing a primary attenuation on the diffraction multiple wave data in the second original data to obtain second primary attenuation data;
[0128] Inputting the detection data into the prediction model to obtain the region where the residual diffraction multiple waves exist in the detection data output by the prediction model, wherein the detection data includes the second primary attenuation data;
[0129] The residual diffraction multiple wave data in the region where the diffraction multiple waves exist are subjected to secondary attenuation to obtain the primary wave data.
[0130] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is further provided, wherein the storage medium stores at least one executable instruction, and when the executable instruction is executed on a computing device, the computing device executes the diffraction multiple wave attenuation method in any of the above method embodiments.
[0131] The executable instructions may be used to cause the computing device to perform the following operations:
[0132] Performing a primary attenuation on the diffraction multiple wave data in the first original data to obtain first primary attenuation data;
[0133] Inputting training data into a neural network for training to obtain a prediction model for predicting a region where residual diffraction multiple waves exist, the training data including first primary attenuation data having a residual diffraction multiple wave region label;
[0134] Performing a primary attenuation on the diffraction multiple wave data in the second original data to obtain second primary attenuation data;
[0135] Inputting the detection data into the prediction model to obtain the region where the residual diffraction multiple waves exist in the detection data output by the prediction model, wherein the detection data includes the second primary attenuation data;
[0136] The residual diffraction multiple wave data in the region where the diffraction multiple waves exist are subjected to secondary attenuation to obtain the primary wave data.
[0137] The algorithm or display provided herein is not inherently related to any particular computer, virtual system or other equipment. Various general purpose systems can also be used together with the teachings based on this. According to the above description, it is obvious to construct the structure required for this type of system. In addition, the embodiment of the present invention is not directed to any specific programming language yet. It should be understood that various programming languages can be utilized to realize the content of the present invention described herein, and the description made to specific languages above is for disclosing the best mode of the present invention.
[0138] In the description provided herein, a large number of specific details are described. However, it is understood that embodiments of the present invention can be practiced without these specific details. In some instances, well-known methods, structures and techniques are not shown in detail so as not to obscure the understanding of this description.
[0139] Similarly, it should be understood that in order to streamline the present invention and aid in understanding one or more of the various inventive aspects, in the above description of exemplary embodiments of the present invention, various features of the embodiments of the present invention are sometimes grouped together into a single embodiment, figure, or description thereof. However, this method of disclosure should not be interpreted as reflecting an intention that the claimed invention requires more features than those expressly recited in each claim.
[0140] Those skilled in the art will appreciate that the modules in the devices in the embodiments may be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or components in the embodiments may be combined into one module or unit or component, and may be divided into a plurality of submodules or subunits or subcomponents. Except that at least some of such features and / or processes or units are mutually exclusive, all features disclosed in this specification (including the accompanying claims, abstracts and drawings) and all processes or units of any method or device disclosed in this manner may be combined in any combination. Unless otherwise expressly stated, each feature disclosed in this specification (including the accompanying claims, abstracts and drawings) may be replaced by an alternative feature providing the same, equivalent or similar purpose.
[0141] It should be noted that the above embodiments illustrate the present invention rather than limit it, and that those skilled in the art may devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference symbol between brackets shall not be construed as a limitation on the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "one" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present invention may be implemented by means of hardware comprising a number of different elements and by means of a suitably programmed computer. In a unit claim enumerating a number of devices, several of these devices may be embodied by the same hardware item. The use of the words first, second, and third, etc. does not indicate any order. These words may be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be understood as limitations on the order of execution.
Claims
1. A diffraction multiple wave attenuation method, characterized in that: include: Performing a primary attenuation on the diffraction multiple wave data in the first original data to obtain first primary attenuation data; Inputting training data into a neural network for training to obtain a prediction model for predicting the area where residual diffraction multiple waves exist, wherein the training data includes the first primary attenuation data having a residual diffraction multiple wave area label; Performing a primary attenuation on the diffraction multiple wave data in the second original data to obtain second primary attenuation data; Inputting detection data into the prediction model to obtain the residual diffraction multiple wave existence area in the detection data output by the prediction model, wherein the detection data includes the second primary attenuation data; The residual diffraction multiple wave data in the residual diffraction multiple wave existing area is subjected to secondary attenuation to obtain primary wave data.
2. The diffraction multiple wave attenuation method according to claim 1, characterized in that: The step of performing a primary attenuation on the diffraction multiple wave data in the first original data to obtain first primary attenuation data includes: Predicting the diffraction multiple wave data present in the first original data to obtain first predicted diffraction multiple wave data; subtracting the first predicted diffraction multiple wave data from the first original data to obtain the first initial attenuation data; and / or The first attenuation of the diffraction multiple wave data in the second original data is performed to obtain the second first attenuation data, including Predicting the diffraction multiple wave data present in the second original data to obtain second predicted diffraction multiple wave data; The second predicted diffraction multiple wave data is subtracted from the second original data to obtain the second primary attenuation data.
3. The diffraction multiple wave attenuation method according to claim 2, characterized in that: After the first predicted diffraction multiple wave data is subtracted from the first original data to obtain the first initial attenuation data, the method further includes: Performing reflected wave attenuation on the first predicted diffraction multiple wave data and / or the first initial attenuation data to obtain first predicted diffraction multiple wave data with reflected waves removed and / or first initial attenuation data with reflected waves removed; The training data is input into the neural network for training to obtain a prediction model for predicting the area where the residual diffraction multiple waves exist, wherein the training data also includes the first predicted diffraction multiple wave data of the reflected wave removed with the residual diffraction multiple wave area label and / or the first initial attenuation data of the reflected wave removed with the residual diffraction multiple wave area label.
4. The diffraction multiple wave attenuation method according to claim 2, characterized in that: After the second predicted diffraction multiple wave data is subtracted from the second original data to obtain the second initial attenuation data, the method further comprises: performing reflected wave attenuation on the second predicted diffraction multiple wave data and / or the second initial attenuation data to obtain second predicted diffraction multiple wave data with reflected waves removed and / or second initial attenuation data with reflected waves removed; The detection data is input into the prediction model to obtain an area where the residual diffraction multiple waves exist in the detection data output by the prediction model, wherein the detection data also includes the second predicted diffraction multiple wave data after removing the reflected wave and / or the second initial attenuation data after removing the reflected wave.
5. The diffraction multiple wave attenuation method according to any one of claims 1 to 4, characterized in that: The step of performing secondary attenuation on the residual diffraction multiple wave data in the region where the residual diffraction multiple waves exist to obtain primary wave data comprises: Acquire an offset distance corresponding to the detection data, where the offset distance is a distance between a signal transmitting device and a signal receiving device that acquires the second raw data; Determine the window width according to the offset; The secondary attenuation is performed on the residual diffraction multiple wave data according to the determined window width.
6. The diffraction multiple wave attenuation method according to claim 5, characterized in that: The number of the detection data is multiple, and the multiple detection data correspond to at least two offset distances of different sizes; The determining the window width according to the offset distance comprises: Obtain a first window width corresponding to a maximum offset and a second window width corresponding to a minimum offset; The window width corresponding to the offset between the maximum offset and the minimum offset is determined by linear interpolation.
7. The diffraction multiple wave attenuation method according to any one of claims 2 to 4, characterized in that: The step of performing secondary attenuation on the residual diffraction multiple wave data in the region where the residual diffraction multiple waves exist to obtain primary wave data comprises: learning diffraction multiple wave characteristics in the first predicted diffraction multiple wave data; Using the learned diffraction multiple wave characteristics, a diffraction multiple wave model is established according to the residual diffraction multiple wave data; The diffraction multiple wave model is subtracted from the residual diffraction multiple wave data to obtain the primary wave data.
8. A diffraction multiple wave attenuation device, characterized in that: include: A first initial attenuation module, used for performing initial attenuation on the diffraction multiple wave data in the first original data to obtain first initial attenuation data; A training module, used for inputting training data into a neural network for training to obtain a prediction model for predicting the area where residual diffraction multiple waves exist, wherein the training data includes the first primary attenuation data having a residual diffraction multiple wave area label; A second initial attenuation module, used for performing initial attenuation on the diffraction multiple wave data in the second original data to obtain second initial attenuation data; A prediction module, used for inputting detection data into the prediction model to obtain the residual diffraction multiple wave existence area in the detection data output by the prediction model, wherein the detection data includes the second initial attenuation data; The secondary attenuation module is used to perform secondary attenuation on the residual diffraction multiple wave data in the area where the residual diffraction multiple waves exist to obtain primary wave data.
9. A computing device, characterized in that include: A processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute the diffraction multiple wave attenuation method according to any one of claims 1 to 7.
10. A computer storage medium, characterized in that: The storage medium stores at least one executable instruction, and the executable instruction enables the processor to execute the diffraction multiple wave attenuation method according to any one of claims 1 to 7.
