A method and system for designing a compressed sensing seismic acquisition geometry

Abstract

The application provides a compressed sensing seismic acquisition observation system design method and system, and belongs to the technical field of oilfield development and seismic data acquisition. The method uses a seismic dictionary of a work area to perform sparse representation on a signal, then optimizes the position distribution of a geophone and the position distribution of a shot point, and obtains an optimized irregular acquisition observation system. The application combines a dictionary learning method and a compressed sensing theory, obtains an optimal irregular observation system under the premise of meeting an exploration task, effectively reduces the number of geophones and shot points, reduces acquisition cost, maximally acquires underground information through limited cost, and has important significance for economic and efficient seismic acquisition work.

Description

Technical Field

[0001] This invention belongs to the field of seismic data acquisition technology, specifically relating to a design method and system for a compressed sensing seismic acquisition and observation system. Background Technology

[0002] With the widespread adoption of "two-width, one-height" seismic exploration technology, the amount of data collected in seismic exploration has increased dramatically, leading to increasingly higher costs for seismic acquisition and thus limiting its widespread application. Therefore, reducing acquisition costs to achieve the desired imaging accuracy using "two-width, one-height" seismic technology has become a pressing need.

[0003] The prerequisite for compressed sensing reconstruction is that the signal possesses sparse characteristics. The ability to achieve the optimal sparse representation of the original signal significantly impacts the final reconstruction quality, and the sparse representation depends on the choice of the transform basis. Therefore, finding a basis function that enables the optimal sparse representation of the original signal is crucial. The earliest sparse transforms employed were traditional orthogonal transforms, such as the Fourier transform, discrete cosine transform, Gabor transform, and wavelet transform. However, the Fourier transform lacks time localization capabilities and cannot represent the spectral characteristics of local regions for non-stationary signals. While the Gabor transform can provide localized video features, it remains a fixed-resolution analysis. The wavelet transform cannot optimally represent high-dimensional signals containing linear or surface singularities. Later, sparse transform methods evolved to multi-scale geometric analysis, using rectangular "strokes" of different directions and lengths to approximate curves, thus improving upon the shortcomings of the wavelet transform. Multi-scale geometric analysis methods include ridge waves, curve waves, and contour waves.

[0004] Existing technologies primarily rely on traditional fixed basis function types to sparsely represent signals, and then optimize the design of observation systems based on this. However, when faced with complex and massive seismic data, it is difficult to achieve satisfactory sparse representation using fixed basis functions. Therefore, compressed sensing acquisition using fixed basis function methods rarely yields good results.

[0005] Chinese patent publication CN109407143A discloses a design method for an irregular seismic exploration observation system based on compressed sensing. This method includes: Step 1, determining the work area, the range of shot points and receiver points, and the required number of shot points, receiver lines, and receiver points; Step 2, designing receiver lines, optimizing their layout based on the work area and the number of receiver lines; Step 3, designing receiver points, optimizing their layout based on the determined receiver line locations; Step 4, designing shot points, optimizing their layout based on the shot point layout range; and Step 5, generating an optimized irregular seismic exploration observation system based on compressed sensing, using the optimized shot point and receiver point locations. This method minimizes reconstruction signal errors through system optimization, providing accurate data for refined reservoir prediction and oil and gas exploration. However, the method disclosed in this patent document uses fixed basis functions for sparse representation, which is difficult to adapt to complex seismic data. Summary of the Invention

[0006] The purpose of this invention is to solve the problems existing in the prior art and provide a design method and system for a compressed sensing seismic acquisition and observation system. By optimizing the design of the observation system, the possibility of recovering the acquired signal into a high-density acquisition signal is greater.

[0007] This invention is achieved through the following technical solution:

[0008] In a first aspect, the present invention provides a design method for a compressed sensing seismic acquisition and observation system. The method utilizes a seismic dictionary for the work area to perform sparse representation of the signal, and then optimizes the location distribution of the geophones and the location distribution of the shot points to obtain an optimized irregular acquisition and observation system.

[0009] A further improvement of the present invention is that:

[0010] The method includes:

[0011] (1) Determine the range of shot points and receiving points and the size of the observation template based on the work area and exploration tasks, and design a dense, regular acquisition observation system;

[0012] (2) Determine the total number of detectors and the total number of shots based on the densely rule-based observation system, reconstruction conditions, and cost;

[0013] (3) Optimize the location distribution of the geophones based on the seismic data dictionary of the work area, the scope of the work area and the total number of geophones to obtain the optimal geophone location;

[0014] (4) Optimize the location distribution of shot points based on the seismic data dictionary of the work area, the scope of the work area and the total number of shots to obtain the optimal shot point location;

[0015] (5) Within the work area, an irregular acquisition and observation system based on compressed sensing is formed by the optimal geophone position and the optimal shot position.

[0016] A further improvement of the present invention is that:

[0017] The operation of step (3) includes:

[0018] (31) A seismic data dictionary applicable to the work area is obtained by training the common shot point gathers based on the simulated data or actual data of the work area;

[0019] (32) Determine the number of detectors in each detector line based on the total number of detectors;

[0020] (33) Based on the seismic data dictionary of the work area and the number of geophones in each geophone line, obtain the optimized geophone observation matrix;

[0021] (34) The positions of each detector within the same detector line are obtained based on the optimized detector observation matrix, i.e. the optimal detector positions.

[0022] A further improvement of the present invention is that:

[0023] If the detector line spacing remains constant, and the detector spacing within the same detector line changes randomly, then the optimal detector position can be obtained using steps (31) to (34).

[0024] A further improvement of the present invention is that:

[0025] If the spacing between detector lines varies randomly, and the spacing between detectors within the same detector line remains constant, then one detector line can be considered as one point, and Nl detector lines are equivalent to Nl detectors within the same detector line.

[0026] Then, using steps (31) to (34), the positions of N1 detectors are obtained, i.e., the optimal detector line positions;

[0027] Finally, the position of each detector in each detector line is obtained by the observation system based on the dense rules collected in step (1), that is, the optimal detector position.

[0028] A further improvement of the present invention is that:

[0029] If the spacing between detector lines changes randomly, and the spacing between detectors within the same detector line changes randomly, then first consider one detector line as a point, and N1 detector lines are equivalent to N1 detectors within the same detector line; then use steps (31) to (34) to obtain the positions of N1 detectors, i.e., the optimal detector line positions.

[0030] Then, by using steps (31) to (34), the position of each detector in each detector line is obtained, thus obtaining the optimal detector position.

[0031] A further improvement of the present invention is that:

[0032] The operation of step (4) includes:

[0033] (41) A seismic data dictionary applicable to the work area is obtained by training the common receiver point gathers based on the simulated data or actual data of the work area;

[0034] (42) Determine the number of firing points in each firing line based on the total number of firing points;

[0035] (43) Based on the seismic data dictionary of the work area and the number of shot points in each shot line, obtain the optimized shot point observation matrix;

[0036] (44) The positions of each shot point within the same shot line are obtained based on the optimized shot point observation matrix, i.e. the optimal shot point positions.

[0037] A further improvement of the present invention is that:

[0038] If the distance between the firing lines remains constant, and the distance between the firing points within the same firing line changes randomly, then the optimal firing point position can be obtained by using steps (41) to (44).

[0039] A further improvement of the present invention is that:

[0040] If the spacing between gun lines changes randomly, and the spacing between gun points within the same gun line remains constant, then one gun line is considered as one point, and N2 gun lines are equivalent to N2 gun points within the same gun line.

[0041] Then, using steps (41) to (44), the positions of N2 gun points are obtained, which are the optimal gun line positions;

[0042] Finally, the position of each shot point within each shot line is obtained from the observation system based on the dense rules collected in step (1), i.e., the optimal shot point position.

[0043] A further improvement of the present invention is that:

[0044] If the distance between the shot lines changes randomly, and the distance between the detectors in the same shot changes randomly, then first consider one shot line as one point, and N2 shot lines are equivalent to N2 shot points in the same shot line; then use steps (41) to (44) to obtain the positions of N2 shot points, that is, the optimal shot line position;

[0045] Then, by using steps (41) to (44), the positions of each gun point within each gun line are obtained, thus obtaining the optimal gun point position.

[0046] A second aspect of the present invention provides a design system for a compressed sensing seismic acquisition and observation system, the system comprising:

[0047] Regular observation system design unit: used to determine the range of shot points and receiver points and the size of observation templates based on the work area and exploration tasks, and to design a dense, regular acquisition observation system;

[0048] Quantity determination unit: Connected to the regular observation system design unit, it is used to determine the total number of detectors and the total number of shots based on the densely acquired regular observation system, reconstruction conditions, and cost.

[0049] Detector location optimization unit: connected to the quantity determination unit, used to optimize the location distribution of detectors based on the seismic data dictionary of the work area, the scope of the work area and the total number of detectors, and obtain the optimal detector location;

[0050] Shot point location optimization unit: connected to the quantity determination unit, used to optimize the location distribution of shot points based on the seismic data dictionary of the work area, the scope of the work area and the total number of shots, and to obtain the optimal shot point location;

[0051] Irregular acquisition system acquisition unit: connected to the detector position optimization unit and the shot point position optimization unit respectively, used to construct an irregular acquisition and observation system based on compressed sensing within the work area, using the optimal detector position and the optimal shot point position.

[0052] A third aspect of the present invention provides a computer-readable storage medium storing at least one computer-executable program, which, when executed by the computer, causes the computer to perform the steps in the above-described compressed sensing seismic acquisition and observation system design method.

[0053] Compared with the prior art, the beneficial effects of the present invention are:

[0054] This invention combines dictionary learning methods and compressed sensing theory to obtain an optimal irregular observation system while meeting exploration requirements. This effectively reduces the number of detectors and shot points, lowers acquisition costs, and maximizes the acquisition of underground information with limited resources. This is of great significance for economical and efficient seismic acquisition work. Attached Figure Description

[0055] Figure 1 A flowchart illustrating the steps of the method of this invention;

[0056] Figure 2 This is a flowchart of the dictionary learning method in the embodiment;

[0057] Figure 3 This is a schematic diagram showing the locations of the dense and regular detector points designed in the embodiment;

[0058] Figure 4 This is a schematic diagram of the optimized sparse and irregular detector point locations in the embodiment. Detailed Implementation

[0059] The present invention will now be described in further detail with reference to the accompanying drawings:

[0060] Seismic acquisition and observation systems describe the relative spatial relationships between shot points and geophones (a one-to-one correspondence between receiver points and geophones) in seismic exploration.

[0061] Both traditional orthogonal transforms and multi-scale geometric analysis use fixed basis function types, thus limiting their expressive power when dealing with diverse signal types. Building on this, researchers have proposed adaptive sparse representation methods to achieve better sparse representation results. Subsequently, researchers extended orthogonal basis functions to redundant dictionaries, leading to the development of numerous dictionary learning methods, including the Optimal Direction Method (MOD) with Expectation Maximization, the k-Singular Value Decomposition (K-SVD) algorithm, and the Iterative Least Squares-based Dictionary Learning Algorithm (ILS-DLA).

[0062] Sparse transform determines the sparsity of a signal. Due to the importance of sparse transform, it is crucial to choose a sparse basis appropriately to minimize the number of non-zero coefficients obtained after the sparse transform. This not only improves sampling and compression speed and reduces transmission bandwidth but also saves storage space, making subsequent signal reconstruction simpler, easier, and more accurate.

[0063] This invention utilizes a dictionary learning algorithm to primarily address the sparse representation problem of signals. The dictionary obtained through learning from a large amount of data can represent seismic data in a more sparse manner, thus minimizing the number of non-zero coefficients obtained after the sparse transformation of the signal.

[0064] The purpose of this invention is to provide a design method for irregular seismic exploration observation systems based on dictionary learning-based compressed sensing theory and simulated annealing optimization algorithm. This invention uses an iterative least squares dictionary learning algorithm to sparsely represent signals using a seismic dictionary for the work area (described in steps 3.1.1 and 4.1.1 below; solving for the adaptive dictionary involves training to obtain the adaptive dictionary). Then, the location distribution of the geophones and shot points is optimized to obtain an optimized irregular acquisition and observation system. Signals acquired using this system have a higher probability of being recovered as high-density acquisition signals.

[0065] The embodiments of the method of the present invention are as follows:

[0066] Example 1

[0067] This invention provides a design method for a compressed sensing seismic acquisition and observation system based on dictionary learning, the method comprising the following steps:

[0068] (1) Determine the range of shot points and receiver points, and the size of the observation template based on the work area and exploration tasks, and design a dense, regular acquisition observation system. This step is implemented using existing methods and will not be elaborated here.

[0069] (2) Determine the total number of detectors and shots based on the densely patterned observation system, reconstruction conditions, and cost. This step is implemented using existing methods and will not be elaborated further here.

[0070] (3) Optimize the location distribution of the geophones based on the seismic data dictionary of the work area, the scope of the work area and the number of geophones to obtain the optimal geophone location.

[0071] (4) Optimize the location distribution of shot points based on the seismic data dictionary of the work area, the scope of the work area and the total number of shots to obtain the optimal shot point location.

[0072] (5) Within the work area, an irregular acquisition and observation system based on compressed sensing is formed by the optimal geophone position and the optimal shot position.

[0073] In step (3), the detectors can be distributed in the following three ways:

[0074] ① The detector line spacing remains constant, while the detector spacing (channel spacing) within the same detector line varies randomly;

[0075] ② The detector line spacing varies randomly, while the detector spacing (channel spacing) within the same detector line remains constant;

[0076] ③ The spacing between detector lines varies randomly, and the spacing between detectors (channel spacing) within the same detector line varies randomly.

[0077] Correspondingly, the operation of optimizing the location distribution of detectors is also divided into three types, including:

[0078] Step 3.1, assuming the detector line spacing remains constant, the detector spacing (channel spacing) within the same detector line varies randomly, i.e., optimizing the positions of Nr detectors within the same line, includes the following steps:

[0079] Step (31) involves training a seismic data dictionary D suitable for the work area based on the common shot point gathers of the simulated or actual data of the work area (the seismic data dictionary D is obtained using existing methods, which will not be elaborated here). In this way, the seismic data can be represented as Y = DX, where X is a sparse coefficient vector.

[0080] The dictionary D and the sparse coefficient vector X are obtained through an iterative least squares dictionary learning algorithm. The formula is:

[0081]

[0082] Step (32) determines the number of detectors in each detector line based on the total number of detectors in step (2). This step uses existing technology to determine the number of detectors in each detector line, and will not be described in detail here.

[0083] Step (33), design the detector observation matrix:

[0084] Based on the seismic data dictionary of the work area obtained in step 3.1.1 and the number of detectors in each detector line, an optimized observation matrix Φ is obtained through simulated annealing algorithm (the optimized observation matrix is ​​implemented using the existing simulated annealing algorithm, which will not be described in detail here). This optimized observation matrix Φ minimizes the correlation of the sensing matrix Ψ.

[0085] The sensing matrix Ψ = ΦD. Minimizing the correlation of the sensing matrix Ψ means minimizing the maximum cross-correlation value between any two column vectors of the sensing matrix. The maximum cross-correlation value between the column vectors is:

[0086] The detector observation matrix Φ shows the positions of Nr detectors within the same line. The specific optimization process is as follows:

[0087] First, a detector observation matrix Φ is given (the detector observation matrix only contains the positions of each detector; the design method is existing technology and will not be elaborated here). A Cor is calculated. Finally, a detector observation matrix Φ is found through simulated annealing algorithm. The Cor value calculated using this detector observation matrix is ​​the minimum. Thus, the detector observation matrix Φ corresponding to the minimum maximum cross-correlation value Cor is the optimized detector observation matrix.

[0088] In step (34), the optimized detector observation matrix Φ gives the positions of Nr detectors within the same detector line. Therefore, the positions of Nr detectors within the same detector line are determined according to the optimized detector observation matrix Φ, thus obtaining the optimal detector positions.

[0089] Step 3.2: Assuming the detector line spacing changes randomly, and the detector spacing (channel spacing) within the same detector line remains constant, the optimal detector position is obtained by optimizing the positions of different detector lines. The specific steps are similar to (31) to (34). At this time, a detector line can be regarded as a point, so N1 detector lines are equivalent to N1 detectors within the same detector line (the value of N1 is obtained through step (32)). The positions of N1 detectors are obtained using the same method as (31)-(34), which is the optimal position of N1 detector lines, or the optimal detector line position. Since the detectors within each detector line are equally spaced, the position of each detector within each detector line can be determined according to the dense rule acquisition observation system in step 1, thus obtaining the optimal detector position. Generally, N1 < <Nr。

[0090] Step 3.3, assuming the detector line spacing varies randomly, and the detector spacing (channel spacing) within the same detector line also varies randomly, this step is performed in two steps:

[0091] Step 3.3.1: Optimize the positions of different detector lines. The specific method is the same as in step 3.2. This step yields the optimal detector line positions.

[0092] Step 3.3.2: Optimize the positions of Nr detectors within the same detector line. The specific steps are the same as in 3.1. This step obtains the positions of each detector in each detector line determined in 3.3.1, thus obtaining the optimal detector positions.

[0093] In step (4), the distribution of shot points can be divided into the following three types:

[0094] ① The distance between the firing points remains constant, while the distance between firing points within the same firing line varies randomly;

[0095] ② The spacing between the shot points varies randomly, while the spacing between shot points within the same shot line remains constant;

[0096] ③ The spacing between shot points varies randomly, and the spacing between shot points within the same shot line also varies randomly.

[0097] Correspondingly, the operation of optimizing the location distribution of firing points is also divided into three types, including:

[0098] Step 4.1, assuming the spacing between shot points remains constant, and the spacing between shot points within the same shot line varies randomly, that is, optimizing the positions of Ns shot points within the same shot line to obtain the optimal shot point positions, includes the following steps:

[0099] (41) Based on the common receiver gathers of the simulated or actual data of the work area, a seismic data dictionary D suitable for the area is obtained. In this way, the seismic data can be represented as Y = DX, where X is a sparse coefficient vector.

[0100] (42) Determine the number of firing points within each firing line based on the total number of firing points in step (2). The number of firing points within each firing line is determined using existing technology, which will not be elaborated here.

[0101] (43) Design the shot point observation matrix.

[0102] Based on the seismic data dictionary D obtained from dictionary learning and the number of shot points in each shot line, the observation matrix Φ is optimized using the simulated annealing algorithm to minimize the correlation of the sensing matrix Ψ.

[0103] Where the sensing matrix Ψ = ΦD, the minimum correlation of the sensing matrix Ψ is achieved by minimizing the maximum cross-correlation value between any two column vectors of the sensing matrix. The maximum cross-correlation value between the column vectors is:

[0104] The shot observation matrix Φ shows the positions of all shot points within the same shot line. The specific optimization steps are as follows:

[0105] Given a shot observation matrix Φ (containing only the positions of each shot point), a Cor is calculated. Using a simulated annealing algorithm, a final shot observation matrix Φ is found that minimizes the Cor value calculated from the shot observation matrix. The shot observation matrix Φ corresponding to the minimum maximum cross-correlation value Cor is the optimized shot observation matrix. The optimized shot observation matrix Φ provides the positions of Ns shot points within the same shot line.

[0106] (44) The optimized shot observation matrix Φ gives the positions of Ns shot points within the same shot line. Therefore, the positions of Ns shot points within the same shot line are determined according to the optimized shot observation matrix Φ, thus obtaining the optimal shot point positions.

[0107] Step 4.2: Assuming the shot line spacing changes randomly, and the shot point spacing within the same shot line remains constant, the positions of different shot lines are optimized. The specific steps are similar to (41)-(44). At this time, a shot line can be regarded as a point, so N2 shot lines are equivalent to N2 shot points within the same shot line (N2 is obtained through step 4.1.2). The positions of N2 shot points are obtained using the same method as (41)-(44), which is the optimal position of N2 shot lines, or the optimal shot line position. Since the detectors within each shot line are equally spaced, the position of each shot point within each shot line can be determined according to the dense rule acquisition observation system in step 1, thus obtaining the optimal shot point position. Generally, N2 < <Ns。

[0108] Step 4.3, assuming the spacing between gun lines varies randomly, and the spacing between gun points within the same gun line also varies randomly, this step is performed in two steps:

[0109] Step 4.3.1: Optimize the positions of different firing lines. The specific steps are the same as in step 4.2. This step obtains the positions of each firing line.

[0110] Step 4.3.2: Optimize the positions of Ns gun points within the same gun line. The specific steps are the same as in 4.1. This step obtains the positions of each gun point in each gun line determined in 4.3.1, that is, the optimal gun point positions are obtained.

[0111] The data used in step (4) and step (3) are different. Step (3) uses the common shot point gather, while step (4) uses the common detector gather. Moreover, the total number of detectors and the total number of shot points are very different. At the same time, the distance between detectors is small, while the distance between shot points is large. Therefore, the calculation results of the two are completely different.

[0112] This invention automatically learns the characteristics of seismic data in the work area using a dictionary learning method to obtain an optimal dictionary (i.e., the work area seismic data dictionary D), thereby obtaining the optimal sparse representation of the seismic data (Y = DX), providing a foundation for the design of the observation system. Based on this, according to the principle of compressed sensing, a simulated annealing global optimization algorithm is used to optimize the positions of the geophones and shot points respectively, minimizing the maximum correlation of the sensing matrix. Ultimately, an optimal irregular observation system is obtained, maximizing the acquisition of subsurface information at a limited cost, ensuring the effectiveness of subsequent seismic data reconstruction, and providing an effective approach for economical and efficient acquisition in "two-width, one-height" seismic exploration. Furthermore, this invention uses a dictionary learning method to adaptively update the dictionary, employing different dictionaries for sparse representation, resulting in stronger adaptability to complex seismic data.

[0113] Examples of applying the method of the present invention are as follows:

[0114]

Example 2

[0115] like Figure 1 As shown, Figure 1 The flowchart below shows the design method of the compressed sensing seismic acquisition and observation system based on dictionary learning according to the present invention.

[0116] Step 001: Based on the work area and exploration tasks, determine the range of shot points and receiver points, the total number of shots, and the size of the observation template, and design a dense, regular acquisition observation system. A schematic diagram of the dense, regular receiver point locations designed in this embodiment is shown below. Figure 3 As shown.

[0117] Step 002: Determine the number of detectors and the total number of shots based on the densely rule-based observation system, reconstruction conditions, and cost.

[0118] Step 003: Optimize the location distribution of the detectors based on the work area and the number of detectors.

[0119] Step 003-1: As an example, the optimization process is described using one of the sparse methods. Here, it is assumed that the detector line spacing remains constant, while the detector spacing (channel spacing) within the same line changes randomly, that is, optimizing the positions of Nr detectors within the same line.

[0120] Step 003-2: Train a dictionary D suitable for the region based on the simulated data and actual data of the work area. In this way, the seismic data Y can be represented as Y = DX, where X is a sparse coefficient vector. Figure 2 The diagram shows the steps involved in training the existing dictionary.

[0121] Step 003-3: Observation Matrix Design. Based on the optimal dictionary obtained through dictionary learning, the observation matrix Φ is optimized using simulated annealing to minimize the correlation of the sensing matrix Ψ, thus obtaining the optimal irregular observation system. Here, the sensing matrix Ψ = ΦD. Minimizing the correlation of the sensing matrix Ψ means minimizing the maximum cross-correlation value between any two column vectors of the sensing matrix. The maximum cross-correlation value between the column vectors is:

[0122] Step 004: Optimize the location distribution of firing points based on the work area and the total number of firing points. Since the principle is similar to Step 003, it will not be described in detail here.

[0123] Step 005: Within the work area, based on the optimized detector and shot point locations, a non-irregular acquisition and observation system based on compressed sensing is formed. A schematic diagram of the optimized sparse, irregular detector point locations in the embodiment is shown below. Figure 4 As shown.

[0124] The present invention also provides a design system for a compressed sensing seismic acquisition and observation system, and an embodiment of the system is as follows:

[0125]

Example 3

[0126] The system includes:

[0127] Regular observation system design unit: used to determine the range of shot points and receiver points and the size of observation templates based on the work area and exploration tasks, and to design a dense, regular acquisition observation system;

[0128] Quantity determination unit: Connected to the regular observation system design unit, it is used to determine the total number of detectors and the total number of shots based on the densely acquired regular observation system, reconstruction conditions, and cost.

[0129] Detector location optimization unit: connected to the quantity determination unit, used to optimize the location distribution of detectors based on the seismic data dictionary of the work area, the scope of the work area and the total number of detectors, and obtain the optimal detector location;

[0130] Shot point location optimization unit: connected to the quantity determination unit, used to optimize the location distribution of shot points based on the seismic data dictionary of the work area, the scope of the work area and the total number of shots, and to obtain the optimal shot point location;

[0131] Irregular acquisition system acquisition unit: connected to the detector position optimization unit and the shot point position optimization unit respectively, used to construct an irregular acquisition and observation system based on compressed sensing within the work area, using the optimal detector position and the optimal shot point position.

[0132] The present invention also provides a computer-readable storage medium, embodiments of which are as follows:

[0133]

Example 4

[0134] The computer-readable storage medium stores at least one computer-executable program, which, when executed by the computer, causes the computer to perform the steps in the above-described compressed sensing seismic acquisition and observation system design method.

[0135] This invention utilizes dictionary learning methods and compressed sensing theory to design an irregular sparse acquisition and observation system. It considers the influence of the inherent characteristics of seismic data on the observation matrix (as reflected in steps 3 and 4, where the dictionary obtained from the seismic data contains the inherent characteristics of the seismic data, and the design of the observation matrix is ​​directly affected by the dictionary). It designs multiple sparsity methods and performs random sampling that meets the requirements of compressed sensing signal reconstruction. This achieves the design of an irregular sparse observation system that meets geological tasks with minimal sampling, and can achieve seismic exploration objectives to the greatest extent and at the lowest cost.

[0136] This invention utilizes a dictionary learning method to adaptively learn the features of seismic data, enabling the seismic data to obtain an optimal sparse representation. Then, combined with compressed sensing theory, the sensing matrix is ​​optimized based on the optimal dictionary to acquire the collected information with the highest probability, resulting in an optimized sparse observation matrix (as shown in steps 3.1.3 and 4.1.3). Based on the observation matrix, the shot-receiver point locations are designed, ultimately resulting in an irregular seismic acquisition and observation system. While preserving the original seismic information to the greatest extent, this system reduces the configuration of receiver and shot points, thereby reducing field acquisition costs and providing an acquisition foundation for the large-scale promotion of costly "two-wide-one-high" seismic exploration.

[0137] In summary, this invention combines dictionary learning methods and compressed sensing theory. The dictionary learning method can adaptively learn seismic data features to achieve a sparser representation. Combined with compressed sensing theory, the location distribution of detectors and shot points is optimized respectively, resulting in an optimal irregular observation system. This system maximizes the acquisition of subsurface information at a limited cost, ensuring the effectiveness of subsequent seismic data reconstruction.

[0138] Finally, it should be noted that the above technical solution is only one embodiment of the present invention. For those skilled in the art, based on the application methods and principles disclosed in the present invention, it is easy to make various types of improvements or modifications, and not limited to the methods described in the above specific embodiments of the present invention. Therefore, the methods described above are only preferred and have no limiting significance.

Claims

1. A design method for a compressed sensing seismic acquisition and observation system, characterized in that: The method comprises the following steps: Step 1, determining the range of shot points and receiving points, and the size of an observation template according to the range of a work area and exploration tasks, and designing a dense regular acquisition observation system; Step 2, determining the total number of geophones and the total number of shots according to the dense regular acquisition observation system, reconstruction conditions and costs; Step 3, optimizing the position distribution of geophones according to a work area seismic data dictionary, the range of the work area and the total number of geophones, and obtaining optimal geophone positions; The operation of the step 3 comprises: Step 31, training a work area seismic data dictionary suitable for the region according to common shot point gathers of simulation data or actual data of the work area; Step 32, determining the number of geophones in each geophone line according to the total number of geophones; Step 33, obtaining an optimized geophone observation matrix according to the work area seismic data dictionary and the number of geophones in each geophone line; Step 34, obtaining the positions of geophones in the same geophone line, i.e. optimal geophone positions, according to the optimized geophone observation matrix; If the geophone line spacing is constant and the geophone spacing in the same geophone line is randomly changed, the optimal geophone positions are obtained by using steps 31 to 34; If the geophone line spacing is randomly changed and the geophone spacing in the same geophone line is constant, one geophone line is regarded as a point, and N1 geophone lines are equivalent to N1 geophones in the same geophone line; then the positions of the N1 geophones, i.e. optimal geophone line positions, are obtained by using steps 31 to 34; finally, the positions of geophones in each geophone line, i.e. optimal geophone positions, are obtained according to the dense regular acquisition observation system in step 1; If the geophone line spacing is randomly changed and the geophone spacing in the same geophone line is randomly changed, one geophone line is regarded as a point, and N1 geophone lines are equivalent to N1 geophones in the same geophone line; then the positions of the N1 geophones, i.e. optimal geophone line positions, are obtained by using steps 31 to 34; and the positions of geophones in each geophone line, i.e. optimal geophone positions, are obtained by using steps 31 to 34; Step 4, optimizing the position distribution of shot points according to the work area seismic data dictionary, the range of the work area and the total number of shots, and obtaining optimal shot point positions; The operation of the step 4 comprises: Step 41, training a work area seismic data dictionary suitable for the region according to common geophone gathers of simulation data or actual data of the work area; Step 42, determining the number of shot points in each shot line according to the total number of shots; Step 43, obtaining an optimized shot point observation matrix according to the work area seismic data dictionary and the number of shot points in each shot line; Step 44, obtaining the positions of shot points in the same shot line, i.e. optimal shot point positions, according to the optimized shot point observation matrix; If the shot line spacing is constant and the shot spacing in the same shot line is randomly changed, the optimal shot point positions are obtained by using steps 41 to 44; If the shot line interval is randomly changed and the shot point interval in the same shot line is constant, a shot line is regarded as a point, and N2 shot lines are equivalent to N2 shot points in the same shot line. Then, the positions of the N2 shot points, i.e., the optimal shot line positions, are obtained by using steps 41 to 44. Finally, the positions of the shot points in each shot line, i.e., the optimal shot point positions, are obtained according to the densely and regularly collected observation system in step 1. If the shot line interval is randomly changed and the receiver interval in the same shot line is randomly changed, a shot line is regarded as a point, and N2 shot lines are equivalent to N2 shot points in the same shot line. Then, the positions of the N2 shot points, i.e., the optimal shot line positions, are obtained by using steps 41 to 44. Then, the positions of the shot points in each shot line, i.e., the optimal shot point positions, are obtained by using steps 41 to 44. Step 5, in the working area range, the optimal receiver positions and the optimal shot point positions are used to form the non-regularly collected observation system based on compressed sensing.

2. A system for designing a compressed sensing seismic acquisition geometry based on the method of claim 1, characterized in that: The system comprises: a regular observation system design unit configured to determine the range of shot points and receiving points, the size of an observation template, and design a densely and regularly collected observation system according to the working area range and the exploration task; a quantity determination unit connected with the regular observation system design unit and configured to determine the total number of receivers and the total number of shots according to the densely and regularly collected observation system, the reconstruction condition, and the cost; a receiver position optimization unit connected with the quantity determination unit and configured to optimize the distribution of receiver positions according to the working area seismic data dictionary, the working area range, and the total number of receivers, and obtain the optimal receiver positions; a shot point position optimization unit connected with the quantity determination unit and configured to optimize the distribution of shot point positions according to the working area seismic data dictionary, the working area range, and the total number of shots, and obtain the optimal shot point positions; a non-regularly collected system acquisition unit connected with the receiver position optimization unit and the shot point position optimization unit respectively, and configured to form a non-regularly collected observation system based on compressed sensing in the working area range by using the optimal receiver positions and the optimal shot point positions.

3. A computer-readable storage medium, characterized in that: The computer readable storage medium stores at least one computer executable program, and the at least one program is executed by the computer to make the computer execute the steps in the compressed sensing seismic acquisition observation system design method according to claim 1.

Images