Seismic record noise removal method and system, electronic equipment and storage medium
Through supervised deep learning methods and Unet network structure, the problem of removing abnormal environmental noise in earthquake records is solved, and the signal-to-noise ratio is improved and the protection of opposite waves and first-to-point waves is achieved.
Patent Information
- Application Number
- CN202311868886.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-12-29
- Publication Date
- 2025-07-01
AI Technical Summary
The prior art is difficult to effectively remove abnormal ambient noise in earthquake recordings, especially at first arrival and surface waves, which often lead to misjudgment and signal-to-noise ratio reduction.
The supervised deep learning method is adopted to use the Unet network structure to record the environmental noise during the acquisition process, create a data set, and train it to remove abnormal environmental noise. The optimized Unet network structure is designed, the encoder-decoder network structure is used, and the loss function and the Adam algorithm are defined using the L1 norm for parameter updates.
Effectively remove abnormal environmental noise in earthquake recording, improve signal-to-noise ratio, and reduce damage to opposite waves and first-to-first waves.
Smart Images
Figure CN120233419A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of seismic record processing in oil exploration, and particularly relates to a method, a system, an electronic device and a storage medium for removing noise from seismic records. Background Art
[0002] Currently, during the acquisition process of seismic record data, abnormal noise caused by environmental interference will reduce the signal-to-noise ratio of seismic records and affect the accuracy of seismic records. In order to avoid the trouble caused by abnormal environmental noise to subsequent data processing, corresponding denoising methods need to be adopted to remove the environmental noise.
[0003] Currently, whether it is various traditional transform-domain denoising methods, such as adaptive filtering, curvelet transform, wavelet transform, etc., or deep learning-based denoising methods, they mainly aim to remove random noise or coherent noise. Random noise or coherent noise has strong regularity and usually weak energy. For abnormal environmental noise, the commonly used method currently is to scan the energy of the seismic profile, set a threshold to determine whether there is abnormal amplitude, and then give an energy attenuation coefficient to directly suppress the abnormal amplitude or even clear the abnormal trace. This denoising method relying on energy scanning often produces misjudgments when encountering the first arrivals, surface waves of the seismic profile, or data near the shot point. Therefore, when using traditional methods, we will limit the processing to only the noise below the first arrivals to prevent this denoising method from damaging the first arrival data.
[0004] The energy scanning denoising method has inherent defects. This method will judge valid mutation signals as noise, and because it is for contiguous abnormal noise, a large window is required to identify it, while a small window is required for single-trace anomalies, resulting in difficulty in choosing the size of the threshold window. In recent years, with the development of machine learning methods, many scholars have applied machine learning to seismic record processing and interpretation work and achieved good results. Compared with traditional denoising methods, deep learning-based denoising methods do not require too many prior assumptions and can utilize a large amount of data sets to deeply mine the features contained in seismic records, thereby achieving the purpose of suppressing noise and improving the signal-to-noise ratio of data.
[0005] Deep learning methods can generally be divided into supervised learning, unsupervised learning, and self-supervised learning according to the training method. Supervised learning denoising methods usually have better effects, but the difficulty lies in the production of labels. Although unsupervised or self-supervised learning avoids label production, their denoising types are limited and they cannot handle noise with complex features. Therefore, the present invention uses the environmental noise records above the first arrivals of seismic records and proposes a method for directly removing abnormal environmental noise from seismic records through supervised deep learning in the time domain.
[0006] The above technical problems need to be solved urgently. Summary of the Invention
[0007] To solve the above technical problems, the present invention proposes a method, system, electronic device and storage medium for removing seismic recording noise to solve the above technical problems.
[0008] The first aspect of the present invention discloses a method for removing seismic recording noise, the method comprising:
[0009] Step S1, data preprocessing: collect environmental noise records above the first arrival and seismic records below the first arrival;
[0010] Step S2, making a data set: make slices of a noise data set and slices of a seismic data set without strong noise, and by randomly combining the slices of the noise data set and the slices of the seismic data set without strong noise, use them as the data set;
[0011] Step S3, removing seismic recording noise: design and optimize the Unet network structure, perform model training for removing seismic recording noise, and after the network training is completed, input the seismic record to be denoised for actual application.
[0012] According to the method of the first aspect of the present invention, in step S2, the method of making slices of a noise data set and slices of a seismic data set without strong noise includes making slices of a noise data set according to the environmental noise records above the first arrival and making slices of a seismic data set without strong noise according to the seismic records below the first arrival.
[0013] According to the method of the first aspect of the present invention, in step S2, the method of making the slices of the noise data set according to the environmental noise records above the first arrival and making the slices of the seismic data set without strong noise according to the seismic records below the first arrival includes:
[0014] Step S21, extracting environmental noise records: calculate the first arrival time according to the given surface velocity and source-receiver relationship, and subtract a small time from the first arrival time to ensure that all records above the first arrival are environmental noise records;
[0015] Step S22, making slices of the noise data set: identify strong noise channels in the environmental noise records above the first arrival, set the weak noise channels to zero values to obtain slices of the noise data set;
[0016] Step S23, slices of the seismic data set without strong noise: identify strong noise channels in the seismic records below the first arrival, and remove the strong noise channels to obtain slices of the seismic data set without strong noise.
[0017] According to the method of the first aspect of the present invention, in step S22, the method of identifying strong noise channels includes: when the following formula holds, the channel where the A tsm value is located is a strong noise channel: A tsm >aA nm ; A tsmis the absolute mean value of a time slice of each trace in the noise part, A nm is the absolute mean value of the noise part, and a is the threshold coefficient for determining the noise intensity.
[0018] According to the method of the first aspect of the present invention, in step S23, the method for identifying strong noise traces includes: when the following formula holds, A tsm the trace where the value is located is a strong noise trace: A tsm > bA nm , b ≤ a; A tsm is the absolute mean value of a time slice of each trace in the noise part, A nm the absolute mean value of the noise part, a is the threshold coefficient for determining the noise intensity, and b is the judgment threshold of the noise when processing the seismic signal part.
[0019] According to the method of the first aspect of the present invention, in step S3, the Unet network structure adopts an encoder-decoder network structure.
[0020] According to the method of the first aspect of the present invention, in step S3, the L1 norm is used to define the loss function of the network, the gradients of each parameter are calculated by backpropagation, and the parameters are updated through the Adam algorithm to minimize the loss function. When the value of the loss function no longer decreases, the training ends.
[0021] The second aspect of the present invention discloses a system for removing seismic recording noise. The system includes:
[0022] The first processing module is configured to perform data preprocessing: collect environmental noise records above the first arrival and seismic records below the first arrival;
[0023] The second processing module is configured to make a data set: make slices of the noise data set and slices of the seismic data set without strong noise, and randomly combine the slices of the noise data set and the slices of the seismic data set without strong noise as the data set;
[0024] The third processing module is configured to remove seismic recording noise: design and optimize the Unet network structure, perform model training for removing seismic recording noise, and after the network training is completed, input the seismic record to be denoised for actual application.
[0025] The third aspect of the present invention discloses an electronic device. The electronic device includes a memory and a processor. When the processor executes a computer program, it implements the steps in any one of the methods for removing seismic recording noise in the first aspect of the present disclosure.
[0026] The fourth aspect of the present invention discloses a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in any one of the methods for removing noise from seismic records in the first aspect of the present disclosure are implemented.
[0027] In summary, the solution proposed in the present invention can use the environmental noise in the process of seismic record acquisition to produce a data set, and train the data set through the optimized Unet network. The application effect of actual data shows that this method can effectively remove abnormal environmental noise in seismic records, improve the signal-to-noise ratio of seismic records, and has less damage to surface waves and first arrival waves. BRIEF DESCRIPTION OF THE DRAWINGS
[0028] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0029] Figure 1 is a flow chart of a method for removing noise from seismic records according to an embodiment of the present invention;
[0030] Figure 2 A schematic diagram of a data set preparation process according to an embodiment of the present invention;
[0031] Figure 3 A schematic diagram of the Unet network structure according to an embodiment of the present invention;
[0032] Figure 4 A schematic diagram showing changes in the loss function and the signal-to-noise ratio with the number of iterations according to an embodiment of the present invention;
[0033] Figure 5 This is a diagram showing the effect of seismic records before denoising in a denoising application according to an embodiment of the present invention;
[0034] Figure 6 This is a diagram showing the effect of denoising seismic records in a denoising application according to an embodiment of the present invention;
[0035] Figure 7 A diagram showing the effect of noise removal in a denoising application according to an embodiment of the present invention;
[0036] Figure 8 is a structural diagram of a system for removing noise from seismic records according to an embodiment of the present invention; and
[0037] Figure 9 The figure is a structural diagram of an electronic device according to an embodiment of the present invention. Detailed implementation manners
[0038] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.
[0039] The first aspect of the present invention discloses a method for removing seismic recording noise. Figure 1 As shown in the flowchart of a method for removing seismic recording noise according to an embodiment of the present invention, as Figure 1 shown, the method includes:[[]]
[0040] Step S1, data preprocessing: Collect environmental noise records above the first arrival and seismic records below the first arrival;
[0041] Step S2, making a data set: Make slices of a noise data set and slices of a seismic data set without strong noise, and use the randomly combined slices of the noise data set and the slices of the seismic data set without strong noise as the data set;
[0042] Step S3, removing seismic recording noise: Design an optimized Unet network structure, perform model training for removing seismic recording noise, and after the network training is completed, input the seismic record to be denoised for actual application. In step S1, data preprocessing: Collect environmental noise records above the first arrival and seismic records below the first arrival.
[0043] In some embodiments, in step S1, for the collected seismic records, it is necessary to obtain environmental noise records above the first arrival and obtain seismic records below the first arrival.
[0044] In step S2, making a data set: Make slices of a noise data set and slices of a seismic data set without strong noise, and use the randomly combined slices of the noise data set and the slices of the seismic data set without strong noise as the data set.
[0045] In some embodiments, in step S2, the method for making slices of a noise data set and slices of a seismic data set without strong noise includes making slices of a noise data set according to the environmental noise records above the first arrival and making slices of a seismic data set without strong noise according to the seismic records below the first arrival.
[0046] In some embodiments, in step S2, the method for making slices of a noise data set according to the environmental noise records above the first arrival and making slices of a seismic data set without strong noise according to the seismic records below the first arrival includes:[[]]
[0047] Step S21, Extract environmental noise records: Calculate the first arrival time according to the given surface velocity and source-receiver relationship, and subtract a small time from the first arrival time to ensure that all records above the first arrival are environmental noise records;
[0048] Step S22, Produce noise dataset slices: In the environmental noise records above the first arrival, identify strong noise channels, set weak noise channels to zero values, and obtain noise dataset slices;
[0049] Step S23, Seismic dataset slices without strong noise: In the seismic records below the first arrival, identify strong noise channels, remove the strong noise channels, and obtain seismic dataset slices without strong noise.
[0050] Specifically, Figure 2 is a schematic diagram of the dataset production process according to an embodiment of the present invention. As Figure 2 shown, for the collected seismic records, it is necessary to obtain the environmental noise records above the first arrival to produce the dataset. Calculate the first arrival time according to the given surface velocity and source-receiver relationship, and then subtract a small time dt from the calculated first arrival time to ensure that all records above this time are environmental noise records. Statistically analyze the extracted environmental noise records, identify strong noise channels, and set weak noise channels to zero values to produce noise dataset slices; for the seismic records below the first arrival, remove the part containing strong noise channels to produce seismic dataset slices without strong noise. Screen out the parts of the seismic slices that contain a large number of empty channels and only retain the slices with fewer empty channels. Randomly combine the noise dataset slices and the seismic dataset slices without strong noise as the dataset data input to the network. Use the seismic dataset slices without strong noise as the label data for training.
[0051] In some embodiments, in step S22, the method for identifying strong noise channels includes: when the following formula holds, the channel where the A tsm value is located is a strong noise channel: A tsm > aA nm ; where A tsm is the absolute mean value of a period slice of each channel in the noise part, A nm is the absolute mean value of the noise part, and a is the threshold coefficient for determining the noise intensity.
[0052] In some embodiments, in step S23, the method for identifying strong noise channels includes: when the following formula holds, the channel where the A tsm value is located is a strong noise channel: A tsm > bA nm , b ≤ a; where A tsm is the absolute mean value of a period slice of each channel in the noise part, A nmis the absolute mean of the noise part, a is the threshold coefficient for determining the noise intensity, and b is the noise judgment threshold when processing the seismic signal part.
[0053] In step S3, seismic record noise removal: Design the Unet network structure, train and optimize the model for seismic record noise removal, and input the data that needs to be denoised for actual application.
[0054] In some embodiments, in step S3, the Unet network structure adopts an encoder-decoder network structure.
[0055] Specifically, Figure 3 is the schematic diagram of the Unet network structure according to the embodiment of the present invention. As Figure 3 shown, the entire network design adopts the Unet network structure, making full use of the convolutional neural network to extract features from the data. Except for the upsampling, downsampling, and output layers, the ReLU is set as the activation function for the remaining convolutional layers. The model adopts an "encoder-decoder" structure. The encoder performs four downsamplings in total, using a convolutional layer with a stride of 2 and a kernel size of 2 to reduce the spatial resolution. The decoder correspondingly performs four upsamplings, using a transposed convolution with the same stride and kernel size for upsampling. There is a skip connection between the encoder and the decoder, with a convolutional layer and a residual block in the middle. Except for upsampling and downsampling, the kernel size of the remaining convolutional layers is 3.
[0056] In some embodiments, in step S3, the L1 norm is used to define the loss function of the network, the gradients of each parameter are calculated by backpropagation, and the parameters are updated through the Adam algorithm to minimize the loss function. When the value of the loss function no longer decreases, the training ends.
[0057] Specifically, in the process of solving the optimized network, the L1 norm is used to define the loss function of the network:
[0058] loss=||seis label -seis pre ||1
[0059] In the formula, seis label represents the label data of the slice of the seismic data set without strong noise, and seis pre represents the data of the slice of the noise data set. According to the loss value calculated by the loss function, the gradients of each parameter are calculated by backpropagation, and the parameters are updated through the Adam algorithm to minimize the loss function. As Figure 4 shown, Figure 4 shows the schematic diagram of the change of the loss function and the signal-to-noise ratio with the number of iterations in the embodiment of the present invention.
[0060] In summary, the solution proposed by the present invention can utilize the environmental noise during the seismic record acquisition process to create a dataset, and train the dataset through the optimized Unet network. The application effect of actual data shows that this method can effectively remove the abnormal environmental noise in the seismic record, improve the signal-to-noise ratio of the seismic record, and cause less damage to surface waves and first-arrival waves.
[0061] Example 1
[0062] 1. Extract data uniformly across the entire work area at a certain ratio, generally extracting at least 30 shot data. The data used to create the dataset needs to include the corresponding environmental noise records.
[0063] 2. Select an appropriate direct wave velocity, calculate the linear first-arrival time, and subtract dt from the calculated linear first-arrival time. By default, dt is 40 ms to ensure that all the data before the linear first arrival is environmental noise.
[0064] 3. According to the calculated linear first-arrival time, calculate the average absolute amplitude of each trace above the first-arrival time. By scanning the amplitude distribution, quickly identify the abnormal noise traces in the seismic record and set the weak noise traces to zero. The judgment criterion is that when the following formula holds, the trace where the A tsm value is located is a strong noise trace. Extract the environmental noise above the linear first arrival after setting the weak noise traces to zero to create noise dataset slices.
[0065] A tsm > aA nm ;
[0066] where A tsm is the absolute mean value of a period slice of each trace in the noise part, A nm is the absolute mean value of the noise part, and a is the threshold coefficient for determining the noise intensity.
[0067] 4. For the corresponding seismic record, set to zero the part containing strong noise traces to create clean seismic data labels. In the seismic record, still use its corresponding noise profile to identify strong noise traces and set them to zero in the seismic record. When the following formula holds, the trace where the A tsm value is located is a strong noise trace. To obtain a relatively cleaner seismic dataset slice without strong noise, the value of b needs to be less than or equal to a.
[0068] A tsm > bA nm , b ≤ a;
[0069] where A tsm is the absolute mean value of a period slice of each trace in the noise part, A nmis the absolute mean of the noise part, a is the threshold coefficient for determining the noise intensity, and b is the noise judgment threshold when processing the seismic signal part.
[0070] 5. Slice the environmental noise records and seismic records respectively to obtain the noise dataset slices and seismic dataset slices without strong noise, and screen the noise dataset slices and seismic dataset slices without strong noise. Remove the parts with all zero values from the noise dataset slices, and at the same time remove the near-offset traces with short environmental noise recording time. For the seismic dataset slices without strong noise, since they contain empty traces, select the slices with the number of empty traces less than c.
[0071] 6. Randomly extract the same number of noise dataset slices and seismic dataset slices without strong noise and combine them as the dataset input to the network. Use the seismic dataset slices without strong noise as the labels of the dataset. Put the dataset into the Unet network constructed in this application for training and optimization. When the loss function value of the dataset no longer decreases and the network converges, the training ends.
[0072] 7. Cut the seismic record data to be denoised into small pieces through slicing technology, then put them into the trained Unet network for prediction to obtain the denoised data slices; rearrange the denoised data slices and merge them into the complete seismic record data, as Figures 5 to 7 shown, Figures 5 to 7 shows the application effect of the actual data, Figure 5 is the pre-denoising seismic record effect diagram in the denoising application according to the embodiment of the present invention, Figure 6 is the post-denoising seismic record effect diagram in the denoising application according to the embodiment of the present invention, Figure 7 is the noise effect diagram removed in the denoising application according to the embodiment of the present invention.
[0073] The second aspect of the present invention discloses a seismic record noise removal system. Figure 8 is the structure diagram of a seismic record noise removal system according to an embodiment of the present invention; as Figure 8 shown, the system 100 includes:
[0074] The first processing module 101 is configured for data preprocessing: collecting environmental noise records above the first arrival and seismic records below the first arrival;
[0075] The second processing module 102 is configured for making a dataset: making noise dataset slices and seismic dataset slices without strong noise, and using the randomly combined noise dataset slices and seismic dataset slices without strong noise as the dataset;
[0076] The third processing module 103 is configured to remove seismic recording noise: design and optimize the Unet network structure, perform model training for seismic recording noise removal, and after the network training is completed, input the seismic recording that needs to be denoised for actual application. The third aspect of the present invention discloses an electronic device. The electronic device includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, the steps in a seismic recording noise removal method according to any one of the first aspects disclosed in the present invention are implemented.
[0077] Figure 9 As shown in the structural diagram of an electronic device according to an embodiment of the present invention, Figure 9 the electronic device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the electronic device is used to communicate with an external terminal in a wired or wireless manner. The wireless manner can be implemented through WIFI, a carrier network, near field communication (NFC), or other technologies. The display screen of the electronic device can be a liquid crystal display screen or an electronic ink display screen. The input device of the electronic device can be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the housing of the electronic device, or an external keyboard, touchpad, or mouse.
[0078] Those skilled in the art can understand that Figure 9 the structure shown in
[0079] is only a structural diagram of a part related to the technical solution of the present disclosure, and does not constitute a limitation on the electronic device to which the solution of the present application is applied. The specific electronic device may include more or fewer components than those shown in the figure, or combine certain components, or have a different component layout.
[0080] Please note that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification. The above embodiments only represent several implementation manners of the present application, and their descriptions are relatively specific and detailed, but they should not be construed as a limitation on the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the appended claims.
[0081] The above are the preferred implementation manners of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can still be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.
Claims
1. A method for removing seismic recording noise, characterized in that, The method includes: Step S1, data preprocessing: Collect environmental noise records above the first arrival and seismic records below the first arrival; Step S2, making a data set: Make slices of the noise data set and slices of the seismic data set without strong noise, and randomly combine the slices of the noise data set and the slices of the seismic data set without strong noise as the training data set; Step S3, seismic record noise removal: Design an optimized Unet network structure, perform model training for seismic record noise removal, and after the network training is completed, input the seismic record to be denoised for actual application.
2. The method for removing seismic recording noise according to claim 1, characterized in that In the step S2, the method for making slices of the noise data set and slices of the seismic data set without strong noise includes making slices of the noise data set according to the environmental noise records above the first arrival, and making slices of the seismic data set without strong noise according to the seismic records below the first arrival.
3. A method for removing seismic recording noise according to claim 2, characterized in that, In the step S2, the method for making slices of the noise data set according to the environmental noise records above the first arrival and slices of the seismic data set without strong noise according to the seismic records below the first arrival includes: Step S21, extracting environmental noise records: Calculate the first arrival time according to the given surface velocity and source-receiver relationship, and subtract a small time from the first arrival time to ensure that all above the first arrival are environmental noise records; Step S22, making slices of the noise data set: In the environmental noise records above the first arrival, identify strong noise channels, set weak noise channels to zero values, and obtain slices of the noise data set; Step S23, slices of the seismic data set without strong noise: In the seismic records below the first arrival, identify strong noise channels, and remove the strong noise channels to obtain slices of the seismic data set without strong noise.
4. The method for removing seismic recording noise according to claim 3, characterized in that, In the step S22, the method for identifying strong noise channels includes: when the following formula holds, the seismic channel where the A tsm value is located is a strong noise channel: A tsm > aA nm ; Among them, A tsm is the absolute mean value of each time slice of the noise part, and A nm is the absolute mean value of the noise part, and a is the threshold coefficient for determining the noise intensity.
5. A method for removing seismic recording noise according to claim 4, characterized in that, In the step S23, the method for identifying strong noise channels includes: when the following formula holds, the seismic channel where the A tsm value is located is a strong noise channel: A tsm > bA nm , b ≤ a; where A tsm is the absolute mean value of a period slice of each channel in the noise part, A nm is the absolute mean value of the noise part, a is the threshold coefficient for determining the noise intensity, and b is the judgment threshold of the noise when processing the seismic signal part.
6. A method for removing seismic recording noise according to claim 5, characterized in that, In the step S3, the Unet network structure adopts an encoder-decoder network structure.
7. The method for removing seismic recording noise according to claim 6, wherein In the step S3, the L1 norm is used to define the loss function of the network, the gradients of each parameter are calculated by backpropagation, and the parameters are updated by the Adam algorithm to minimize the loss function. When the value of the loss function no longer decreases, the training ends.
8. An earthquake recording noise removal system, characterized in that, The system includes: The first processing module is configured to perform data preprocessing: Collect environmental noise records above the first arrival and seismic records below the first arrival; The second processing module is configured to make a data set: Make slices of the noise data set and slices of the seismic data set without strong noise, and randomly combine the slices of the noise data set and the slices of the seismic data set without strong noise as the training data set; The third processing module is configured to perform seismic record noise removal: Design an optimized Unet network structure, perform model training for seismic record noise removal, and after the network training is completed, input the seismic record to be denoised for actual application.
9. An electronic device, characterized in that, The electronic device includes a memory and a processor. When the processor executes the computer program, the steps in any one of claims 1 to 7 of a method for removing seismic record noise are implemented.
10. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium. When the computer program is executed by a processor, the steps in a method for removing seismic recording noise according to any one of claims 1 to 7 are implemented.