A training method for a Fresnel zone picking model and a diffraction wave imaging method based on machine learning

Abstract

The present application provides a Fresnel zone picking model, a training method, and a diffraction wave imaging method based on machine learning. The training method includes obtaining seismic wave training data; annotating Fresnel zone labels for a horizontal slice image set of an initial 3D dip gather to obtain a true label set; inputting the horizontal slice image set of the initial 3D dip gather and the true label set into a pre-constructed Fresnel zone picking model for iterative training to obtain a prediction result of the Fresnel zone; each time during training, updating the parameters of the Fresnel zone picking model based on the loss value between the true label and the prediction result of the Fresnel zone; until a preset iteration condition is reached. The diffraction wave imaging method uses the trained Fresnel zone picking model to perform diffraction wave imaging. The present invention introduces a neural network to pick the Fresnel zone based on a 3D dip gather, so that the picked Fresnel zone has higher accuracy, thereby improving the accuracy and efficiency of diffraction wave imaging.

Description

Technical Field

[0001] The present invention relates to the technical field of geological exploration, and in particular, to a Fresnel zone picking model, a training method, and a diffracted wave imaging method based on machine learning. Background Art

[0002] In geological exploration, the diffracted wave field is generated by small-scale discontinuities underground. Therefore, the diffracted wave contains a large amount of high-resolution information about underground micro-inhomogeneous geological bodies. Therefore, using diffracted wave imaging has natural advantages in identifying small-scale changes in microstructures in geological bodies, can break through the limitation of the main frequency wavelength on the resolution in reflection wave imaging, and is of great significance for improving the identification ability of micro-geological body anomalies in oil and gas exploration.

[0003] The key to diffracted wave imaging is how to separate the diffracted wave from the reflected wave field or how to effectively suppress the reflected wave. Since the main energy of reflected wave imaging comes from the corresponding reflected Fresnel zone, and this reflected Fresnel zone is related to the imaging depth, offset, and formation dip, and shows different forms in different data domains, especially in the area where diffractors exist, it becomes more complex. Therefore, picking the Fresnel zone is a repetitive and very time-consuming task. The prior art uses manual picking, which is inefficient, thus resulting in low imaging efficiency. At the same time, the manual picking results largely depend on the personal experience of the implementer, resulting in low picking accuracy, and thus affecting the imaging accuracy of the diffracted wave. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide a Fresnel zone picking model, a training method, and a diffracted wave imaging method based on machine learning to improve the efficiency and accuracy of diffracted wave imaging.

[0005] In the first aspect, a Fresnel zone picking model is provided, including an encoder and a decoder;

[0006] The encoder includes an input unit and multiple feature extraction units; each feature extraction unit includes a first convolution module, an activation function, and a second convolution module; wherein, the activation function is a Tanh activation function; the convolution kernel size of the second convolution module is 3*3 and the stride is 2;

[0007] The decoder includes feature reconstruction units with the same number as the feature extraction units and an output unit, and each feature reconstruction unit includes a deconvolution module.

[0008] In the second aspect, a training method of a Fresnel zone picking model is provided, and the method includes:

[0009] Obtain seismic wave training data for the target work area; the seismic wave training data includes a horizontal slice image set of the initial 3D dip gathers corresponding to all imaging points on the target imaging line in the seismic wave imaging space;

[0010] Annotate the Fresnel zone labels for the horizontal slice image set of the initial 3D dip gathers to obtain a true label set;

[0011] Input the horizontal slice image set of the initial 3D dip gathers and the true label set into the pre-constructed Fresnel zone picking model for iterative training to obtain the prediction result of the Fresnel zone;

[0012] During each training, update the parameters of the Fresnel zone picking model based on the loss value between the true label and the prediction result of the Fresnel zone; until the preset iteration condition is reached.

[0013] Optionally, annotating the Fresnel zone labels for the horizontal slice image set of the initial 3D dip gathers to obtain a true label set includes:

[0014] Pick the Fresnel zone for each horizontal slice in the horizontal slice image set of the initial 3D dip gathers;

[0015] Excise the picked Fresnel zone;

[0016] Make a true label set according to the horizontal slice image set after excising the Fresnel zone.

[0017] Optionally, the Fresnel zone is elliptical on the horizontal slice; picking the Fresnel zone for each horizontal slice in the horizontal slice image set of the initial 3D dip gathers includes:

[0018] For each horizontal slice in the horizontal slice image set of the initial 3D dip gathers, obtain the 3D dip gather data corresponding to this horizontal slice; the 3D dip gather data includes the two-way travel time of the seismic wave, the x-direction dip, the y-direction dip corresponding to the seismic trace, and the main frequency of the seismic wave;

[0019] Calculate the radii of the Fresnel zone in two directions based on the 3D dip gather data and the picking calculation formula of the Fresnel zone, and determine the range of the Fresnel zone based on the radii in two directions and the stationary phase point; where the picking calculation formula of the Fresnel zone is:

[0020]

[0021] Where, is the radius of the Fresnel zone in the direction; is the radius of the Fresnel zone in the direction; is the two-way travel time of the seismic wave; is the main frequency of the seismic wave; is the stationary phase point of the 3D dip gather at the imaging point on the corresponding horizontal slice, where respectively correspond to the formation dip angles of the formation where the imaging point is located along the x direction and the y direction.

[0022] Optionally, the process of each training during iterative training includes:

[0023] The input unit based on the Fresnel zone picking model receives the horizontal slice image set and the true label set of the initial 3D dip gather;

[0024] Based on the first convolution module, extract the Fresnel zone features of the horizontal slice image set of the initial 3D dip gather, and detect the edges of the Fresnel zones in each horizontal slice based on the Fresnel zone features; among them, the Fresnel zone features at least include the shape, position, and the relative relationship between the energy strengths inside and outside the Fresnel zone.

[0025] Based on the Tanh activation function, increase the complexity of the extracted Fresnel zone features;

[0026] Based on the second convolution module with a convolution kernel size of 3*3 and a stride of 2, reduce the spatial resolution of each slice in the slice image set;

[0027] After multiple feature extractions, gradually restore the spatial resolution of each horizontal slice in the slice image set through the deconvolution module, and fuse the horizontal slices with the same spatial resolution in feature extraction and feature reconstruction;

[0028] Output the prediction result of the finally detected Fresnel zone through the output unit.

[0029] Optionally, the expression of the output unit is:

[0030]

[0031] In the formula, pro represents the probability value that a single pixel point in the output is classified as the Fresnel zone area, which is represented by a value from 0 to 1; m represents the parameter controlling the energy attenuation degree outside the Fresnel zone.

[0032] In a third aspect, a diffraction wave imaging method based on machine learning is provided, and the method includes:

[0033] Obtain the seismic wave detection data of the target work area, and the seismic wave detection data includes the horizontal slice image set of the 3D dip gathers corresponding to all imaging points on all imaging lines in the seismic wave imaging space;

[0034] Input the horizontal slice image set of the 3D dip gather into the pre-trained Fresnel zone picking model to pick the Fresnel zones on the horizontal slices corresponding to each imaging point;

[0035] Remove the picked Fresnel zones from the horizontal slice image set of the 3D dip gather;

[0036] Stack the horizontal slice image set of the 3D dip gather after removing the Fresnel zones to obtain a diffracted wave imaging profile.

[0037] In a fourth aspect, a diffracted wave imaging device based on machine learning is provided. The device includes:

[0038] An acquisition unit for acquiring seismic wave detection data of a target work area, where the seismic wave detection data includes a horizontal slice image set corresponding to all imaging points on all imaging lines in the seismic wave imaging space;

[0039] A picking unit for inputting the horizontal slice image set of the 3D dip gather into a pre-trained Fresnel zone picking model to pick the Fresnel zones on the horizontal slice corresponding to each imaging point;

[0040] A removal unit for removing the picked Fresnel zones from the horizontal slice image set of the 3D dip gather;

[0041] A stacking unit for stacking the horizontal slice image set of the 3D dip gather after removing the Fresnel zones to obtain a diffracted wave imaging profile.

[0042] In a fifth aspect, an electronic device is provided, including a processor, a communication interface, a memory, and a communication bus. Among them, the processor, the communication interface, and the memory complete communication with each other through the communication bus;

[0043] The memory is used to store a computer program;

[0044] The processor is used to implement the method steps described in any one of the second or third aspects when executing the program stored on the memory.

[0045] In a sixth aspect, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the method steps described in any one of the second or third aspects.

[0046] A Fresnel zone picking model, a training method, and a diffraction wave imaging method based on machine learning provided by an embodiment of the present invention are trained based on an improved Fresnel zone picking model. The training method includes obtaining seismic wave training data; annotating Fresnel zone labels for a horizontal slice image set of an initial 3D dip gather to obtain a true label set; inputting the horizontal slice image set of the initial 3D dip gather and the true label set into a pre-constructed Fresnel zone picking model for iterative training to obtain a prediction result of the Fresnel zone; each time during training, updating the parameters of the Fresnel zone picking model based on the loss value between the true label and the prediction result of the Fresnel zone; until a preset iteration condition is reached. And using the trained Fresnel zone picking model to perform diffraction wave imaging. The present invention introduces a neural network to pick the Fresnel zone based on a 3D dip gather. Its horizontal feature is elliptical, which is more in line with the situation of the Fresnel zone in the dip domain compared with the traditional method of extracting the Fresnel zone boundary based on 2D dip gathers in the mutually perpendicular X and Y directions and then forming a rectangle. Therefore, the accuracy of the picked Fresnel zone is higher, which further improves the accuracy of diffraction wave imaging. And by using the trained Fresnel zone picking model to separate the reflected wave and perform diffraction wave imaging, the efficiency is higher.

[0047] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following specific preferred embodiments are given in conjunction with the accompanying drawings and described in detail as follows. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for the embodiments. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0049] Figure 1 Shows a schematic diagram of the network structure of the Fresnel zone picking model provided by the embodiment of the present invention;

[0050] Figure 2 Shows a flowchart of the training method of the Fresnel zone picking model provided by the embodiment of the present invention;

[0051] Figure 3 Shows a schematic diagram of the comparison result before and after removing the Fresnel zone in a three-dimensional space provided by the embodiment of the present invention;

[0052] Figure 4 Shows a schematic diagram of the change of the horizontal slice image set in each layer of the Fresnel zone picking model provided by the embodiment of the present invention;

[0053] Figure 5The figure shows a comparison schematic diagram of the output results before and after the improvement of the output unit provided by the embodiment of the present invention;

[0054] Figure 6 The figure shows a schematic flowchart of the diffraction wave imaging method based on machine learning provided by the embodiment of the present invention;

[0055] Figure 7 The figure shows a comparison schematic diagram before and after removing the Fresnel zone from the horizontal slice corresponding to the imaging point provided by the embodiment of the present invention;

[0056] Figure 8 The figure shows a schematic diagram of the diffraction wave imaging profile provided by the embodiment of the present invention;

[0057] Figure 9 The figure shows a schematic diagram of the reflection wave imaging profile in the prior art;

[0058] Figure 10 The figure shows a schematic structural diagram of the diffraction wave imaging device based on machine learning;

[0059] Figure 11 The figure shows a schematic structural diagram of the electronic device provided by the embodiment of the present invention. Detailed implementation manners

[0060] To make the objectives, technical solutions and advantages of the embodiments 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. Obviously, the described embodiments are only some of the embodiments of the present invention, rather than all of the embodiments. Usually, the components of the embodiments of the present invention described and illustrated herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0061] The principle of using seismic waves for geological exploration is as follows: First, one or more seismic sources and multiple geophones are set above the ground. The seismic waves generated by the multiple seismic sources are transmitted downward to the ground in sequence or simultaneously. After being reflected by each geological surface layer of the ground, they are reflected back to the ground, and the reflected seismic waves are received by all or some of the geophones. Prestack migration processing is performed according to parameters such as the amplitude and travel time of the detected reflected waves to obtain underground complex structure information.

[0062] In the prior art, imaging is mainly carried out by using reflected waves. However, the resolution limit of the reflected wave imaging method is one-fourth of the main frequency wavelength of seismic waves. Even if various frequency domain protection or extension methods are adopted, its resolution still cannot meet the needs of actual production.

[0063] Therefore, in the embodiments of the present invention, diffraction wave imaging is used. The diffraction wave field is generated by small-scale discontinuities underground. Therefore, the diffraction wave contains a large amount of high-resolution information about underground micro-inhomogeneous geological bodies. Therefore, using diffraction waves has natural advantages in identifying small-scale changes in the above-mentioned microstructures, can break through the limitation of the main frequency wavelength on the resolution in reflected wave imaging, and is of great significance for improving the recognition ability of micro-geological body anomalies in oil and gas exploration.

[0064] The key to diffraction wave imaging is to separate the diffraction wave from the reflected wave. The energy of the reflected wave is mainly concentrated in the Fresnel zone area. Therefore, by picking up the Fresnel zone, the area of the reflected wave can be determined, and then separation can be carried out.

[0065] The prior art uses manual picking of the Fresnel zone, with low efficiency, which in turn leads to low imaging efficiency. At the same time, the manual picking results largely depend on the personal experience of the implementer, resulting in low picking accuracy, which in turn affects the imaging accuracy of the diffraction wave.

[0066] Based on this, the embodiments of the present invention provide a Fresnel zone picking model and its training method, which will be described below through embodiments.

[0067] The embodiments of the present invention provide a Fresnel zone picking model, as Figure 1 shown, including an encoder and a decoder;

[0068] The encoder includes an input unit 101 and multiple feature extraction units 102; each feature extraction unit 102 includes a first convolution module, an activation function, and a second convolution module; wherein, the activation function is a Tanh activation function; the convolution kernel size of the second convolution module is 3*3 and the stride is 2.

[0069] The Fresnel zone picking model of the embodiment of the present invention is an improvement based on the traditional U-net network structure. In the traditional U-net network, the activation function of the feature extraction unit is the LeakyReLU activation function, which is not applicable to seismic wave detection data because there are some negative value features in seismic wave detection data. Using the LeakyReLU activation function will cause the negative value features to disappear. Therefore, the embodiment of the present invention uses the Tanh activation function. The Tanh function is continuous and differentiable over the entire real number domain and maps the input to the interval (-1, 1). This means that it can better process and retain the negative value information in seismic wave detection data without the possibility of making negative value features disappear or become less significant as in the case of LeakyReLU. In addition, the Tanh function provides a smoother non-linear transformation than LeakyReLU, which helps the model learn more complex patterns and features, especially when dealing with signal data with a rich dynamic range.

[0070] The feature extraction unit of the traditional U-net network structure uses max pooling to reduce the spatial resolution of the feature map, which will cause some information loss in the feature map. Therefore, the embodiment of the present invention uses a second convolutional module with a convolutional kernel size of 3*3 and a stride of 2 to replace the max pooling layer. Compared with the max pooling operation, using a convolutional module for feature extraction can learn how to best compress the spatial information instead of simply selecting the maximum value of each region, so that more original information can be retained in the downsampled feature map, thereby improving the model performance.

[0071] The decoder includes a feature reconstruction unit 103 and an output unit 104 with the same number as the feature extraction units. Each feature reconstruction unit 103 includes a transposed convolution module.

[0072] The transposed convolution module gradually restores the feature map to its original spatial resolution and splices and fuses it with the feature map of the same spatial resolution in the feature reconstruction unit, and finally outputs through the output unit. The output unit is also improved, which will be described in the following embodiments and will not be elaborated here.

[0073] For the Fresnel zone picking model constructed according to the above embodiments, the embodiment of the present invention provides a training method for the Fresnel zone picking model, as Figure 2 shown, the method includes the following steps:

[0074] Step S201: Obtain seismic wave training data of the target work area.

[0075] The seismic wave training data includes a horizontal slice image set of the initial 3D dip angle gather corresponding to all imaging points on the target imaging line in the seismic wave imaging space.

[0076] In the embodiment of the present invention, the seismic wave imaging space is a three-dimensional imaging space, which includes multiple imaging lines. Each imaging line further includes multiple CDPs (Common Depth Points), and each CDP contains a seismic trace, which includes multiple imaging points.

[0077] In this step, the target imaging line is a relatively typical imaging line. For example, some imaging lines may pass through important geological structures, such as faults, fold axes, salt domes, etc. These areas are crucial for understanding the geological background of the entire exploration area. Therefore, these imaging lines can be determined as target imaging lines. Training with the dip gather data of a small number of typical imaging lines can improve the training accuracy.

[0078] In this step, first, the 3D dip gather corresponding to each imaging point is obtained. This dip gather is obtained after migration processing of the initial prestack gather; since the 3D dip gather is a three-dimensional data, while the Fresnel zone picking model processes a two-dimensional image, it is necessary to perform dimensionality reduction processing on the 3D dip gather to obtain a horizontal slice.

[0079] In a specific example, if the depth of any one imaging point is , then the corresponding 3D dip gather result is:

[0080] (1);

[0081] Where, is the shot point coordinate; is the imaging point coordinate; is the geophone coordinate; is the two-way travel time of the seismic wave; is the travel time of the seismic wave from the seismic source to the imaging point; is the travel time from this imaging point to the geophone; represents taking the first derivative of the seismic trace sequence, is the number of the seismic trace; is the number of seismic traces. is the dip in the is the dip in the

[0082] Where, in the calculation process of the dip gather of this imaging point, for the seismic trace numbered j, its corresponding dip is calculated by the following formula:

[0083] (2);

[0084] (3);

[0085] In the formula, is the root mean square velocity of the imaging point, and the meanings of other parameters refer to those in Formula 1.

[0086] Finally, dimensionality reduction processing is performed on the above-mentioned 3D dip gathers, that is, for a given depth remains unchanged, and thus an image set of horizontal slices corresponding to each depth is obtained.

[0087] Step S202: Label Fresnel zone labels for the image set of horizontal slices of the initial 3D dip gathers to obtain a true label set.

[0088] If all horizontal slices of the imaging space are labeled, it will be a very large workload. Therefore, in order to reduce the workload and improve the picking efficiency, in a feasible implementation manner, labeling Fresnel zone labels for the image set of horizontal slices of the initial 3D dip gathers to obtain a true label set includes:

[0089] Step S2021: Pick the Fresnel zone for each horizontal slice in the image set of horizontal slices of the initial 3D dip gathers.

[0090] When labeling Fresnel zone labels in the training stage, first pick the Fresnel zone by means of human-computer interaction picking. Specifically, the 3D dip gathers are stacked along the x direction and the y direction respectively, and the stationary phase points (i.e., the center points of the Fresnel zones) are picked on the stacked sections in their respective directions. Among them, respectively correspond to the formation dips of the formation where the imaging points are located along the x direction and the y direction. Then, the spatial range of the Fresnel zone is further picked by using the picking results and the Fresnel zone radius calculation formula. Finally, its horizontal slice is performed to obtain the Fresnel zone label.

[0091] In the embodiments of the present invention, the Fresnel zone is elliptical on the horizontal slice; therefore, as long as the major axis radius, minor axis radius and center point of the Fresnel zone are calculated, the position and shape of the Fresnel zone on the horizontal slice can be determined.

[0092] In a feasible implementation manner, picking the Fresnel zone for each horizontal slice in the image set of horizontal slices of the initial 3D dip gathers includes:

[0093] Step S2021A: For each horizontal slice in the image set of horizontal slices of the initial 3D dip gathers, obtain the 3D dip gather data corresponding to this horizontal slice; the 3D dip gather data includes the two-way travel time of the seismic wave, the x-direction dip, y-direction dip corresponding to the seismic trace, and the main frequency of the seismic wave;

[0094] Step S2021B: Calculate the radii of the Fresnel zone in two directions based on the 3D dip gather data and the pick-up calculation formula of the Fresnel zone, and determine the range of the Fresnel zone based on the radii in the two directions and the position of the stationary phase point; where the pick-up calculation formula of the Fresnel zone is:

[0095] (4);

[0096] (5);

[0097] Where, is the radius of the Fresnel zone in the direction; is the radius of the Fresnel zone in the direction; is the two-way travel time of the seismic wave; is the main frequency of the seismic wave; is the stationary phase point of the 3D dip gather at the imaging point on the corresponding horizontal slice, where respectively correspond to the formation dips of the formation where the imaging point is located along the x direction and the y direction.

[0098] Step S2022: Excise the picked-up Fresnel zone.

[0099] In this step, manual excision is adopted.

[0100] Step S2023: Make a true label set according to the horizontal slice image set after excising the Fresnel zone.

[0101] In an example, for instance, label the area of the excised Fresnel zone as 0 and the area outside the Fresnel zone as 1 to form a true label set.

[0102] Figure 3 shows the front and back comparison schematic diagrams of excising the Fresnel zone area in three-dimensional space, that is, in multiple consecutive horizontal slices. The left side is before excision, and the right side is after excision.

[0103] Since the energy of the reflected wave is mainly in the Fresnel zone area, by excising the Fresnel zone, the reflected wave is suppressed, and the separation of the reflected wave and the diffracted wave is achieved.

[0104] Step S203: Input the horizontal slice image set of the initial 3D dip gather and the true label set into the pre-constructed Fresnel zone pick-up model for iterative training to obtain the prediction result of the Fresnel zone.

[0105] Figure 4 shows the change schematic diagrams of the horizontal slice image set of the initial 3D dip gather passing through each layer of the Fresnel zone pick-up model; in this step, the process of each training is elaborated based on the above pre-constructed Fresnel zone pick-up model:

[0106] Step S2031: The input unit based on the Fresnel zone picking model receives the horizontal slice image set and the true label set of the initial 3D dip gather.

[0107] Step S2032: The first convolution module extracts the Fresnel zone features of the horizontal slice image set of the initial 3D dip gather, and detects the edges of the Fresnel zones in each horizontal slice based on the Fresnel zone features.

[0108] In one example, the convolution kernel size of the first convolution module is 3*3 and the stride is 1.

[0109] The Fresnel zone picking model of the embodiment of the present invention has the ability of multi-scale feature extraction. As Figure 4 shown, the shallow network mainly extracts local detail features, and the deep network with a larger receptive field extracts more abstract global features. During the network training process, it can not only fully obtain the macroscopic features of the Fresnel zone, such as shape, position, and the relative relationship of the internal and external energy strengths; but also can identify the detailed morphology of the reflected waves therein, such as frequency, amplitude, and the morphology of the isophase axis, etc. Therefore, this network not only has the ability to determine the position and shape of the Fresnel zone, but also can effectively detect the remaining reflected wave energy outside the Fresnel zone.

[0110] Step S2033: Based on the Tanh activation function, increase the complexity of the extracted Fresnel zone features.

[0111] Step S2034: Based on the second convolution module with a convolution kernel size of 3*3 and a stride of 2, reduce the spatial resolution of each horizontal slice in the slice image set.

[0112] Step S2035: After multiple feature extractions, gradually restore the spatial resolution of each horizontal slice in the slice image set through the deconvolution module, and fuse the horizontal slices with the same spatial resolution in feature extraction and feature reconstruction.

[0113] In this step, more original information can be retained through splicing fusion, and richer detail features can be extracted.

[0114] Step S2036: The output unit outputs the prediction result of the finally detected Fresnel zone.

[0115] In the embodiment of the present invention, since the Fresnel model can effectively detect the remaining reflected wave energy outside the Fresnel zone, in order to make full use of this network to suppress the remaining reflected waves, the output unit is improved.

[0116] The output unit of the traditional U-net network model is a simple binary classification output. Especially at the inner and outer edges of the Fresnel zone, if it is only classified as 0 or 1, high-frequency truncation noise is likely to occur, resulting in an uneven edge.

[0117] The expression of the improved output unit is:

[0118] (6);

[0119] In the formula, represents the probability value that a single pixel in the output is classified as the Fresnel zone area, ranging from 0 to 1; represents the parameter that controls the degree of energy attenuation outside the Fresnel zone.

[0120] This is a formula that converts the original probability value (ranging from 0 to 1) into another form of probability value, where is the control parameter used to adjust the degree of energy attenuation outside the Fresnel zone.

[0121] This conversion method makes the transition near the inner and outer boundaries of the Fresnel zone smoother. By adjusting the parameter the smoothness can be controlled to achieve a better edge processing effect. Such processing is particularly important for improving the quality of the model prediction results, especially in cases where more refined processing is required at the edges or details.

[0122] Specifically, the parameter controls the rate of energy attenuation outside the Fresnel zone. When is larger, the influence of this formula on different values is more significant, resulting in a more rapid change in the output value, especially when is close to 0 or 1. On the contrary, a smaller value will cause the change in the output value to be more gentle, thus achieving a smooth effect at the boundary and reducing the occurrence of high-frequency truncation noise.

[0123] In some cases, appropriately selecting the value can enhance the contrast between the inside and outside of the Fresnel zone, which helps to improve the accuracy of the model in identifying the target area.

[0124] As Figure 5 shown, it is a schematic diagram comparing the output results before and after the improvement of the output unit; the left side is before the improvement, and the right side is after the improvement. It can be clearly observed that the edge of the Fresnel zone after the improvement is smoother.

[0125] Step S204: Each time during training, update the parameters of the Fresnel zone picking model based on the loss value between the true label and the prediction result of the Fresnel zone; until the preset iteration condition is reached.

[0126] In this step, the loss function can adopt the cross-entropy loss function. The preset iteration condition can be reaching the preset number of iterations or the loss value reaching the preset threshold.

[0127] Based on the above embodiments, an embodiment of the present invention provides a diffraction wave imaging method based on machine learning, as Figure 6 shown, the method includes the following steps:

[0128] Step S601: Obtain the seismic wave detection data of the target work area. The seismic wave detection data includes a horizontal slice image set of 3D dip gathers corresponding to all imaging points on all imaging lines in the seismic wave imaging space.

[0129] It should be noted that the target work area is the same as the target work area during training.

[0130] Step S602: Input the horizontal slice image set of the 3D dip gathers into the pre-trained Fresnel zone picking model to pick the Fresnel zone on the horizontal slice corresponding to each imaging point.

[0131] In this step, the specific processing process of the model refers to the above example and will not be elaborated here.

[0132] The model outputs the weight coefficient values for smoothing and removing the Fresnel zone, that is, the probability distribution of the Fresnel zone.

[0133] Step S603: Remove the picked Fresnel zone from the horizontal slice image set of the 3D dip gathers.

[0134] In one example, the removal method can be multiplying the output result of the Fresnel zone picking model by the input, and then the horizontal slice with the Fresnel zone removed can be obtained. Since the area of the output Fresnel zone is set to 0, after multiplying with the original horizontal slice, the corresponding area of the original horizontal slice is also set to 0, thus achieving the removal.

[0135] In this step, the Fresnel zone area can be cut out by multiplying the output result by the input horizontal slice. As Figure 7 shown, it is a comparison schematic diagram before and after the horizontal slice is removed. The left side is before removal, and the right side is after removal.

[0136] Step S604: Stack the horizontal slice image set of the 3D dip gathers after removing the Fresnel zone to obtain a diffraction wave imaging profile.

[0137] In this step, as Figure 8As shown, it is a schematic diagram of the diffraction wave imaging profile obtained by the method of the embodiment of the present invention. To make the effect of the present invention more prominent, it is compared with the traditional reflection wave imaging profile. Figure 9 As shown, it is the reflection wave imaging profile. Figure 8 And Figure 9 For the comparison of the actual data imaging profiles of the same imaging area, it can be seen that there is a good correspondence between the diffraction wave isophase axis in the deep learning diffraction wave imaging result and the disconnection position of the reflection wave isophase axis in the reflection wave imaging result, indicating that the machine learning diffraction wave imaging method based on 3D formation dip gathers proposed by the present invention can effectively suppress the reflection wave energy in seismic data and realize diffraction wave imaging; it can handle the situation of complex underground structures, reflect the spatial position relationship of underground diffractors, and has higher accuracy compared with the reflection wave imaging method.

[0138] Based on the same inventive concept, a diffraction wave imaging device based on machine learning is provided, as Figure 10 shown. The device includes:

[0139] An acquisition unit 1001, configured to acquire seismic wave detection data of a target work area, where the seismic wave detection data includes a horizontal slice image set of 3D dip gathers corresponding to all imaging points on all imaging lines in the seismic wave imaging space.

[0140] A picking-up unit 1002, configured to input the horizontal slice image set of the 3D dip gathers into a pre-trained Fresnel zone picking-up model to pick up the Fresnel zones on the horizontal slices corresponding to each imaging point.

[0141] An excision unit 1003, configured to excise the picked-up Fresnel zones in the horizontal slice image set of the 3D dip gathers.

[0142] A stacking unit 1004, configured to stack the horizontal slice image set of the 3D dip gathers after excising the Fresnel zones to obtain a diffraction wave imaging profile.

[0143] Based on the same technical concept, an embodiment of the present invention further provides an electronic device, as Figure 11 shown, including a processor 1101, a communication interface 1102, a memory 1103, and a communication bus 1104. Among them, the processor 1101, the communication interface 1102, and the memory 1103 communicate with each other through the communication bus 1104.

[0144] The memory 1103 is used to store a computer program;

[0145] The processor 1101, when executing the program stored in the memory 1103, implements the steps of the training method of the Fresnel zone picking-up model and the machine learning-based diffraction wave imaging method.

[0146] The communication bus mentioned in the above electronic device may be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The communication bus can be divided into an address bus, a data bus, a control bus, etc. For the sake of simplicity, only a thick line is used to represent it in the figure, but it does not mean that there is only one bus or one type of bus.

[0147] The communication interface is used for communication between the above electronic device and other devices.

[0148] The memory may include a Random Access Memory (RAM), and may also include a Non-Volatile Memory (NVM), such as at least one disk memory. Optionally, the memory may also be at least one storage device located far from the aforementioned processor.

[0149] The above-mentioned processor may be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0150] The computer program product of the training method for the Fresnel zone pickup model and the diffraction wave imaging method based on machine learning provided by the embodiments of the present invention includes a computer-readable storage medium storing program code, and the instructions included in the program code can be used to execute the methods described in the foregoing method embodiments. For specific implementation, reference can be made to the method embodiments, which will not be elaborated here.

[0151] The device of the diffraction wave imaging method based on machine learning provided by the embodiments of the present invention can be specific hardware on a device, or software or firmware installed on the device, etc. For the device provided by the embodiments of the present invention, its implementation principle and the technical effects produced are the same as those of the foregoing method embodiments. For the sake of brief description, for the parts not mentioned in the device embodiments, reference can be made to the corresponding content in the foregoing method embodiments. Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the foregoing-described systems, devices, and units can all refer to the corresponding processes in the foregoing method embodiments, and will not be elaborated herein.

[0152] In the embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For another example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some communication interfaces. The indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0153] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place, or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0154] In addition, each functional unit in the embodiments provided by the present invention can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0155] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The foregoing storage medium includes: USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical disks and other various media that can store program codes.

[0156] It should be noted that like reference numerals and letters denote like items in the following figures. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are only used for descriptive distinction and should not be construed as indicating or implying relative importance.

[0157] Finally, it should be noted that the above-described embodiments are only specific embodiments of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit it. The protection scope of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: any person skilled in the art within the technical scope disclosed by the present invention can still modify the technical solutions recorded in the foregoing embodiments or can easily think of changes, or make equivalent replacements for some of the technical features; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. All should be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A training method for a Fresnel zone pickup model, characterized in that The method includes: Obtaining seismic wave training data of a target work area; the seismic wave training data includes a horizontal slice image set of an initial 3D dip gather corresponding to all imaging points on a target imaging line in a seismic wave imaging space; Annotating Fresnel zone labels for the horizontal slice image set of the initial 3D dip gather to obtain a true label set; the annotation process is as follows: Picking up the Fresnel zone for each horizontal slice in the horizontal slice image set of the initial 3D dip gather; the Fresnel zone is elliptical on the horizontal slice; the picking-up process is as follows: For each horizontal slice in the horizontal slice image set of the initial 3D dip gather, obtaining 3D dip gather data corresponding to the horizontal slice; the 3D dip gather data includes the two-way travel time of the seismic wave, the dip angle in the x direction, the dip angle in the y direction corresponding to the seismic trace, and the main frequency of the seismic wave; Calculating the radii of the Fresnel zone in two directions based on the 3D dip gather data and the picking-up calculation formula of the Fresnel zone, and determining the range of the Fresnel zone based on the radii in the two directions and the stationary phase point; Removing the picked-up Fresnel zone; Making a true label set according to the horizontal slice image set after removing the Fresnel zone; Inputting the horizontal slice image set of the initial 3D dip gather and the true label set into a pre-constructed Fresnel zone picking-up model for iterative training to obtain a prediction result of the Fresnel zone; During each training, updating the parameters of the Fresnel zone picking-up model based on the loss value between the true label and the prediction result of the Fresnel zone; until a preset iteration condition is reached.

2. The method according to claim 1, characterized in that, The picking-up calculation formula of the Fresnel zone is: Wherein, is the radius of the Fresnel zone in the direction; is the radius of the Fresnel zone in the direction; is the two-way travel time of the seismic wave; is the main frequency of the seismic wave; is the stationary phase point of the 3D dip gather at the imaging point on the corresponding horizontal slice, wherein correspond to the formation dips of the formation where the imaging point is located along the x-direction and y-direction, respectively.

3. The method according to claim 1, characterized in that The process of each training during iterative training includes: Receiving the horizontal slice image set of the initial 3D dip gather and the true label set based on the input unit of the Fresnel zone picking-up model; Extracting the Fresnel zone features of the horizontal slice image set of the initial 3D dip gather based on the first convolution module, and detecting the edges of the Fresnel zone in each slice based on the Fresnel zone features; wherein, the Fresnel zone features at least include the shape, position, and the relative relationship of the energy strength inside and outside the Fresnel zone; Increasing the complexity of the extracted Fresnel zone features based on the Tanh activation function; Reducing the spatial resolution of each horizontal slice in the slice image set based on the second convolution module with a convolution kernel size of 3*3 and a stride of 2; After multiple feature extractions, gradually restoring the spatial resolution of each horizontal slice in the slice image set through a deconvolution module, and fusing the features of the horizontal slices with the same spatial resolution in feature extraction and feature reconstruction; Outputting the prediction result of the finally detected Fresnel zone through the output unit.

4. The method according to claim 3, characterized in that, The expression of the output unit is: In the formula, pro represents the probability value that a single pixel point in the output is classified as the Fresnel zone area, which is represented by a value from 0 to 1; m represents a parameter controlling the energy attenuation degree outside the Fresnel zone.

5. A diffraction wave imaging method based on machine learning, characterized in that, The method includes: Obtaining seismic wave detection data of a target work area, the seismic wave detection data includes a horizontal slice image set of a 3D dip gather corresponding to all imaging points on all imaging lines in a seismic wave imaging space; Input the horizontal slice image set of the 3D dip gather into a pre-trained Fresnel zone picking model to pick the Fresnel zones on the horizontal slices corresponding to each imaging point; wherein, the Fresnel zone picking model is trained by using the method according to any one of claims 1-4; Cut out the picked Fresnel zones in the horizontal slice image set of the 3D dip gather; Stack the horizontal slice image set of the 3D dip gather after cutting out the Fresnel zones to obtain a diffracted wave imaging profile.

6. A diffraction wave imaging device based on machine learning, characterized in that, The device includes: An acquisition unit, configured to acquire seismic wave detection data of a target work area, where the seismic wave detection data includes a horizontal slice image set of a 3D dip gather corresponding to all imaging points on all imaging lines in a seismic wave imaging space; A picking unit, configured to input the horizontal slice image set of the 3D dip gather into a pre-trained Fresnel zone picking model to pick the Fresnel zones on the horizontal slices corresponding to each imaging point; wherein, the Fresnel zone picking model is trained by using the method according to any one of claims 1-4; A cutting unit, configured to cut out the picked Fresnel zones in the horizontal slice image set of the 3D dip gather; A stacking unit, configured to stack the horizontal slice image set of the 3D dip gather after cutting out the Fresnel zones to obtain a diffracted wave imaging profile.

7. An electronic device, characterized in that, Comprising a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; The memory is used for storing a computer program; The processor is configured to, when executing the program stored on the memory, implement the steps of the training method of the Fresnel zone picking model according to any one of claims 1-4 and the diffracted wave imaging method based on machine learning according to claim 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the training method of the Fresnel zone picking model according to any one of claims 1-4 and the diffracted wave imaging method based on machine learning according to claim 5 are implemented.

Images