Stable sandstone type uranium mine seismic data random undersampling method

Abstract

The invention belongs to a sandstone type uranium mine seismic exploration data acquisition technology, and particularly relates to a stable sandstone type uranium mine seismic data random undersampling method, which comprises the following steps of: (1) setting parameters, and determining the number of sampling points reserved after thinning; (2) segmenting all sample points, and determining the number of residual sample points in a middle section; (3) adjusting the number of sample points in the middle section according to the number of residual sample points in the middle section; (4) determining the number of sampling points, and extracting the sampling points; and (5) determining a measurement matrix. According to the method, the problem of reconstruction distortion caused by sample point remaining and sample point missing can be effectively solved, and accurate reconstruction of data can be better realized while the sandstone type uranium mine seismic exploration data acquisition compression ratio is improved.

Description

Technical Field

[0001] The invention belongs to a sandstone-type uranium mine seismic exploration data acquisition technology, and in particular relates to a stable sandstone-type uranium mine seismic data random undersampling method. Background Art

[0002] In recent years, seismic exploration technology, the most accurate geophysical exploration method, has gradually been applied to sandstone uranium exploration. However, there are still many challenges in the practical application of seismic exploration technology for sandstone uranium deposits.

[0003] From an economic perspective, the acquisition construction cost increases significantly with the increase in construction area, blasthole density and operation time. With the continuous deepening of applications, the continuous improvement of acquisition accuracy, the continuous rise in prices, the continuous rise in human resource costs, and the increasing contradictions between workers and farmers, the cost of seismic exploration has also been stimulated to rise steadily. From a technical perspective, the limitations of the surface acquisition environment, coupled with the elimination of waste blastholes and waste blastholes, will objectively lead to irregular and incomplete seismic data. The traditional solution is to change the view, but if the regular sampling grid data is missing on a large scale or the supplementary blastholes are unreasonable, spatial false frequencies will be generated in local blank areas, which will seriously affect the exploration quality and imaging effect of the entire work area, thus bringing difficulties to data processing and interpretation.

[0004] Therefore, how to improve the competitiveness of seismic exploration methods from the perspective of cost and technology is an important issue facing seismic exploration of sandstone-type uranium deposits.

[0005] The establishment and development of compressed sensing theory has made it possible to solve the above problems. Seismic acquisition and processing technology based on compressed sensing theory has been widely used in the fields of oil and gas and coalfield exploration, but has not received much attention in the field of sandstone uranium exploration. The compressed sensing theory can be generally divided into three aspects: compressed measurement, sparse representation, and data reconstruction. Among them, the compressed measurement process is non-adaptive, that is, the measurement matrix cannot change according to the change of the signal, so how to project a high-dimensional signal into a low-dimensional space without losing useful information is the key to the research. Therefore, the measurement matrix is ​​one of the key factors in the compressed sensing theory, and its selection determines whether the signal can be reconstructed in the end.

[0006] In seismic exploration, the construction of measurement matrix is ​​mainly based on random undersampling method. Random undersampling methods include: pure random undersampling, jitter random undersampling, segmented random undersampling, improved segmented random undersampling, and edge-preserving segmented random undersampling. Pure random undersampling, or Gaussian measurement matrix, is essentially a sampling method based on Bernoulli distribution with a probability of 0.5. Its advantage is that it fully satisfies the restricted equidistance property and can theoretically achieve complete reconstruction of the signal, but in practice, the undersampling interval is likely to be too large, resulting in the loss of reconstruction information. Jitter random undersampling can effectively avoid the problem of too large undersampling intervals under the premise of satisfying random distribution, but this method will completely degenerate into regular undersampling when the undersampling ratio (i.e., the ratio of the actual number of sampling points to the designed number of sampling points) is higher than 50%, and no longer meets the requirements of compressed sensing random measurement. Segmented random undersampling is also a method to effectively control the undersampling interval, and there is no problem of degenerating into regular undersampling, but the above methods do not control the number of sample points, and there may be a problem that the remaining sample points cannot be divided evenly. Improved segmented random undersampling completely solves the problem of residual samples. However, since there is no control on whether edge samples are retained, the above methods all have edge distortion, and thus there is a certain reconstruction error at the edge of the profile. Edge-preserving segmented random undersampling solves the problem of edge reconstruction distortion very well. However, it does not specifically consider the problem of residual samples.

[0007] Therefore, it is urgent to propose a random undersampling method for seismic data suitable for sandstone-type uranium deposits, which can effectively solve the problems existing in the above undersampling methods. Summary of the invention

[0008] The purpose of the present invention is to provide a stable random under-sampling method for sandstone-type uranium mine seismic data, which can effectively solve the problem of sample point surplus and reconstruction distortion caused by sample point loss, and can better realize accurate reconstruction of data while improving the compression ratio of sandstone-type uranium mine seismic exploration data acquisition.

[0009] The technical solution to achieve the purpose of the present invention is:

[0010] A stable random undersampling method for seismic data of sandstone-type uranium deposits, the method comprising:

[0011] Step (1) setting parameters to determine the number of sampling points retained after thinning;

[0012] Step (2) segmenting all sample points and determining the number of sample points remaining in the middle segment;

[0013] Step (3) adjusting the number of sample points in the middle segment according to the number of sample points remaining in the middle segment;

[0014] Step (4) determining the number of sampling points and extracting the sampling points;

[0015] Step (5) determines the measurement matrix.

[0016] The sample points are seismic data.

[0017] The step (1) specifically includes: setting the total number of sampling points N, the thinning percentage parameter η and the number of edge-preserving sampling points G, and determining the number of sampling points retained after thinning according to the total number of sampling points N and the thinning percentage parameter η: Indicates rounding.

[0018] In the step (1), the number of edge preservation samples G is 1 to 3, which is much smaller than N.

[0019] The step (2) is specifically as follows: all sampling points are segmented, the number of sampling points in the two edge segments is G, and the number of sampling points in the middle segment is M. r The number of sample points in each segment is Among them, N r Indicates the number of samples in the middle segment, M r Indicates the number of intermediate segments. Indicates rounding; the number of remaining samples in the middle is n r , n r <M r .

[0020] The step (3) is specifically as follows: r Randomly select n r segment, and the remaining n r The sample points are placed one by one into the selected n r In the segment, after adjustment, there will be n r The number of samples in the segment increases by one.

[0021] The step (4) is specifically as follows: r A sample point is randomly selected from each segment as a sampling point, and the middle segment has a total of M r sampling points; all the sampling points in the two edge segments are used as sampling points, and there are 2G sampling points in total in the two edge segments; the total number of sampling points, that is, the number of sampling points retained after thinning is M = M r +2G;

[0022] The step (5) is specifically as follows: the sample points extracted from the sequence, that is, the M sample points retained after thinning, are assigned a value of 1, and are diagonalized, and 0 rows are removed to obtain a measurement matrix A.

[0023] The beneficial technical effects of the present invention are:

[0024] 1. The present invention provides a stable random under-sampling method for sandstone-type uranium mine seismic data. Under the premise of considering the problem of residual sample points, all sample points are divided into two sections, the middle section and the edge section, according to the rarefaction percentage as a unified parameter, the sampling sequence is regrouped and the sample points are allocated in advance. At the same time, a certain number of sample points are reserved at both ends of the edge for edge control. In this way, the unified parameter can simply round off the proportion of the retained sample points for construction design, without the need to understand the specific number of sample points, which is more in line with collection habits and general experience.

[0025] 2. The present invention provides a stable random under-sampling method for sandstone-type uranium mine seismic data. By setting the total number of sample points N, the rarefaction percentage parameter η and the number of edge-maintained sample points G, all sample points are divided into two sections, the middle section and the edge section, the number of sample points remaining in the middle section is determined, and the remaining sample points in the middle section are randomly allocated to the middle section, thereby effectively solving the problem of remaining sample points.

[0026] 3. The present invention provides a stable random undersampling method for sandstone-type uranium mine seismic data, which effectively solves the problem of reconstruction distortion caused by missing sample points by retaining edge holding sample points G at both ends of the edge for edge control.

[0027] 4. The stable random undersampling method for sandstone-type uranium mine seismic data provided by the present invention can better realize the construction of compressed sensing measurement matrix in two-dimensional seismic acquisition and provide robust compressed sampling data for subsequent data reconstruction. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 A flow chart of a stable random under-sampling method for seismic data of sandstone-type uranium deposits provided by the present invention;

[0029] Figure 2 Measurement matrices for different undersampling methods: (a) regular undersampling; (b) pure random undersampling; (c) jitter undersampling; (d) piecewise random undersampling; (e) improved piecewise random undersampling; (f) edge-preserving piecewise random undersampling optimized by the present invention;

[0030] Figure 3 The noise-added undersampling profiles of different undersampling methods are as follows: (a) regular undersampling; (b) pure random undersampling; (c) jitter undersampling; (d) piecewise random undersampling; (e) improved piecewise random undersampling; (f) edge-preserving piecewise random undersampling optimized by the present invention.

[0031] Figure 4 Reconstructed profiles for data of different undersampling methods: (a) regular undersampling; (b) pure random undersampling; (c) jitter undersampling; (d) piecewise random undersampling; (e) improved piecewise random undersampling; (f) edge-preserving piecewise random undersampling optimized by the present invention;

[0032] Figure 5 Reconstruction errors of different undersampling methods: (a) regular undersampling; (b) pure random undersampling; (c) jitter undersampling; (d) piecewise random undersampling; (e) improved piecewise random undersampling; (f) edge-preserving piecewise random undersampling optimized by the present invention. DETAILED DESCRIPTION

[0033] The present invention is further described in detail below with reference to the accompanying drawings and embodiments.

[0034] like Figure 1 As shown, the present invention provides a stable random undersampling method for sandstone-type uranium mine seismic data, which specifically includes the following steps:

[0035] Step (1) Set parameters to determine the number of sampling points to be retained after thinning

[0036] Set the total number of sampling points N, the thinning percentage parameter η and the number of edge-preserving sampling points G.

[0037] According to the total number of sampling points N and the thinning percentage parameter η, the number of sampling points M retained after thinning is determined. Indicates rounding.

[0038] The number of edge-preserving samples G is usually 1 to 3, much smaller than N.

[0039] Step (2) divide all sample points into segments and determine the number of remaining sample points in the middle segment

[0040] Divide all sampling points into segments, with the number of intermediate segments being M r , the number of segments at both ends is 2, and all sampling points are divided into M r +2 segments, that is, the total number of segments is M r +2. The number of sample points in the middle section is N r , the number of sample points in both edge segments is G, and the total number of sample points is N = N r +2G, middle section M r Each paragraph has A sample point, Indicates rounding, at this time the number of remaining sample points in the middle segment is n r , it is obvious that r <M r ;

[0041] Step (3) Adjust the number of sample points in the middle segment according to the number of sample points remaining in the middle segment.

[0042] From the middle section M r Randomly select n r segment, and the remaining n r The sample points are placed one by one into the selected nr In the segment, after adjustment, there will be n r The number of sample points in the segment increases by one;

[0043] Step (4) Determine the number of sampling points and extract the sampling points

[0044] From the middle section M r A sample point is randomly selected from each segment as a sampling point, and the middle segment has a total of M r sampling points; all the sampling points in the two edge segments are used as sampling points, and there are 2G sampling points in total in the two edge segments; the total number of sampling points (i.e. the number of sampling points retained after thinning) M = M r +2G;

[0045] Step (5) Determine the measurement matrix

[0046] The positions of the sample points extracted from the sequence (the M sample points retained after thinning) are assigned a value of 1, and are diagonalized, and the 0 row is removed to obtain the measurement matrix A.

[0047] Each of the above sample points actually represents a piece of seismic data. The above steps are used for random undersampling, which not only solves the problem of too large undersampling intervals, but also eliminates the influence of residual sample points. At the same time, the data edge is maintained, avoiding the problem of too large edge error in data reconstruction.

[0048] Example

[0049] The measurement matrices were measured using different undersampling methods, with a typical value of η = 50%, i.e., half of the seismic data was thinned out by undersampling methods, and the performance of the measurement matrices obtained by different undersampling methods was compared and analyzed. Figure 2 As shown. Figure 2 The results show that the random undersampling method will have significant problems of local over-density sampling and excessive gap spacing, which will inevitably lose small-scale detail information during the reconstruction process, which will be further reflected in the subsequent seismic profiles.

[0050] The forward simulation adopts a simple three-layer model, and 10dB Gaussian noise is added to the forward single shot. Figure 2 The measurement matrix in is used to perform undersampling thinning and data reconstruction using different methods to obtain the noisy profiles and reconstructed profiles of different undersampling methods. The noisy undersampling profiles of different undersampling methods are shown in Figure 2. Figure 3 As shown in Figure 2, the data reconstruction profiles of different undersampling methods are as follows: Figure 4 The reconstruction errors of different undersampling methods are shown in Figure 5 shown.

[0051] from Figure 3-Figure 5 The three groups of figures show significant differences between the thinned and reconstructed sections obtained by different undersampling methods:

[0052] (1) Regular undersampling does not meet the requirements of compressed sensing theory, so the data cannot be accurately reconstructed;

[0053] (2) The pure random undersampling method has the problem of too much missing data due to the large local thinning interval, which leads to the loss of reconstructed local details.

[0054] (3) When η≥50%, jitter undersampling is equivalent to the regular undersampling method, so the same problem as (1) occurs.

[0055] (4) Both the piecewise random undersampling method and the improved piecewise random undersampling method do not consider the problem of edge reconstruction distortion. In addition, the piecewise random undersampling method does not carefully consider the problem of sample point surplus, which leads to the instability of the sampling algorithm.

[0056] (5) The optimized edge-preserving piecewise random undersampling method proposed in the present invention solves the problems of sample point surplus and edge distortion in (4) in the process of compression measurement controlled by the thinning ratio parameter η, thus achieving the best profile reconstruction effect. Figure 5 The minimum reconstruction error is also obtained in the mean error comparison diagram. Therefore, it is the best method among the six undersampling methods for seismic compression measurement.

[0057] The present invention is described in detail above with reference to the accompanying drawings and embodiments, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge of ordinary technicians in the field without departing from the purpose of the present invention. The contents not described in detail in the present invention can adopt the existing technology.

Claims

1. A stable random undersampling method for seismic data of sandstone-type uranium deposits, characterized in that: The method comprises: Step (1) setting parameters to determine the number of sampling points retained after thinning; Step (2) segmenting all sample points and determining the number of sample points remaining in the middle segment; Step (3) adjusting the number of sample points in the middle segment according to the number of sample points remaining in the middle segment; Step (4) determining the number of sampling points and extracting the sampling points; Step (5) determines the measurement matrix.

2. The stable random undersampling method for seismic data of sandstone-type uranium deposits according to claim 1, characterized in that: The sample points are seismic data.

3. The method for random undersampling of seismic data of stable sandstone-type uranium deposits according to claim 1, characterized in that: The step (1) specifically includes: setting the total number of sampling points N, the thinning percentage parameter η and the number of edge-preserving sampling points G, and determining the number of sampling points retained after thinning according to the total number of sampling points N and the thinning percentage parameter η: Indicates rounding.

4. The stable random undersampling method for seismic data of sandstone-type uranium deposits according to claim 3 is characterized in that: In the step (1), the number of edge preservation samples G is 1 to 3, which is much smaller than N.

5. The stable random undersampling method for seismic data of sandstone-type uranium deposits according to claim 3 is characterized in that: The step (2) is specifically as follows: all sampling points are segmented, the number of sampling points in the two edge segments is G, and the number of sampling points in the middle segment is M. r The number of sample points in each segment is Among them, N r Indicates the number of sample points in the middle segment, M r Indicates the number of intermediate segments. Indicates rounding; the number of remaining samples in the middle is n r , n r <M r .

6. A stable random undersampling method for seismic data of sandstone-type uranium deposits according to claim 5, characterized in that: The step (3) is specifically as follows: r Randomly select n r segment, and the remaining n r The sample points are placed one by one into the selected n r In the segment, after adjustment, there will be n r The number of samples in the segment increases by one.

7. A stable random undersampling method for seismic data of sandstone-type uranium deposits according to claim 6, characterized in that: The step (4) is specifically as follows: r A sample point is randomly selected from each segment as a sampling point, and the middle segment has a total of M r sampling points; All the sample points in the two edge segments are used as sampling points, and there are 2G sampling points in total in the two edge segments; The total number of sampling points, that is, the number of sampling points retained after thinning is M = M r +2G.

8. The stable random undersampling method for seismic data of sandstone-type uranium deposits according to claim 7, characterized in that: The step (5) is specifically as follows: the sample points extracted from the sequence, that is, the M sample points retained after thinning, are assigned a value of 1, and are diagonalized, and 0 rows are removed to obtain a measurement matrix A.

Images