CN-121995473-A - Nonlinear constraint time domain high-resolution spectrum inversion method

CN121995473ACN 121995473 ACN121995473 ACN 121995473ACN-121995473-A

Abstract

The invention provides a nonlinear constraint time domain high-resolution spectrum inversion method, which relates to the field of seismic data broadband processing and comprises the steps of constructing a wedge model library composed of odd-even reflection coefficient pairs, obtaining a wedge model library with simple layer response through convolution of the wedge model library and wavelets in the time domain, forming expected inversion seismic data through combination of different layer responses, making difference between the seismic data and expected results, applying an L2-Cauchy constraint design objective function, solving a regularization constraint nonlinear problem by using a rapid iteration shrinkage threshold algorithm to obtain a high-resolution reflection coefficient profile, and obtaining the high-resolution broadband spectrum inversion seismic data through convolution of the high-resolution reflection coefficient profile and broadband wavelets. The method has the beneficial effects that the high-resolution broadband inversion seismic data is obtained by convolving and shaping the broadband wavelet of the high-resolution reflection coefficient profile, the low frequency and the high frequency of the seismic data are compensated, and the thin layer information is effectively identified.

Inventors

WEI GUOHUA
LIU HAOJIE
KONG QINGFENG
LUO PINGPING
QIAN ZHI
SUN WEIGUO
DING KUN

Assignees

中国石油化工股份有限公司
中国石油化工股份有限公司胜利油田分公司

Dates

Publication Date: 20260508
Application Date: 20241104

Claims (10)

1. A non-linear constrained time domain high resolution spectral inversion method, comprising: constructing a wedge-shaped model library consisting of parity reflection coefficient pairs; in the time domain, the odd-even wedge model library and the wavelet convolution are carried out to obtain a wedge model library with simple layer response; Constructing expected inversion seismic data by the combination of different layer responses; making differences between the seismic data and the expected result, and designing an objective function by applying regularization constraint; Solving a regularization constraint nonlinear problem by using a rapid iteration shrinkage threshold algorithm to obtain a high-resolution reflection coefficient profile; And convolving the broadband wavelet with the high-resolution reflection coefficient profile to obtain high-resolution broadband spectrum inversion seismic data.
2. The nonlinear-constrained time-domain high-resolution spectral inversion method according to claim 1, wherein the construction step of the wedge-shaped model library composed of parity-reflection coefficient pairs is as follows: The total reflection coefficient pairs are represented separately, written as even reflection coefficient pairs and odd reflection coefficient pairs, respectively: Where n refers to the number of samples between the two reflectors, t refers to the point in time where the reflective layer is located, r o refers to the odd reflection coefficient pair component, Δt refers to the horizon thickness, and r e refers to the even reflection coefficient pair component; Scaling the even reflection coefficient pairs and the odd reflection coefficient pairs and bringing the two parts together represents: ar e +br 0 =(a+b)δ(t)+(a-b)δ(t+nΔt) (2) wherein a and b are scaling parameters of the parity-check reflection coefficient pair component; any pair of reflection coefficients is decomposed into a sum of a pair of even reflection coefficients and a pair of odd reflection coefficients in a certain proportion, and the pair of even reflection coefficients and the pair of odd reflection coefficients can be combined into any pair of reflection coefficients through a and b, and the pair of reflection coefficients is as follows: r(t,n,Δt)=cδ(t)+dδ(t+nΔt)=ar e +br 0 (3) wherein c= (a+b), d= (a-b); by introducing a wedge matrix, equation (3) is symbolized by a matrix, which is:
3. The non-linearly constrained time-domain high-resolution spectral inversion method of claim 2 wherein for a simple layer model, only one of the pair of scaling parameters a i and b i is non-zero and the remaining scaling parameters should be zero, re-representing equation (4) as a matrix: where U is an even reflection coefficient wedge matrix, a is a scaling parameter sequence of even reflection coefficient pair components, V is an odd reflection coefficient wedge matrix, and b is a scaling parameter sequence of odd reflection coefficient pair components.
4. A non-linear constrained time domain high resolution spectral inversion method according to claim 3 wherein said simple layer response parity wedge model library is constructed by the steps of: The wavelet convolution for each row of equation (5) is: The equation (6) is re-expressed as a matrix: Where s is the constructed layer response; is the convolution of the wavelet and U; is the convolution of the wavelet with V.
5. The nonlinear-constrained time domain high resolution spectral inversion method of claim 1 wherein the expected inversion seismic data is constructed by a combination of different layer responses, the construction steps being as follows: each parity-wedge reflection coefficient pair can be written separately as: where m is the number of samples from the first data to the seismic trace, n is the number of samples between the two reflectors, t is the time point where the reflective layer is located, r o is the odd reflection coefficient pair component, Δt is the horizon thickness, and r e is the even reflection coefficient pair component; the two parts are brought together: Wherein c= (a+b), d= (a-b), Is shift data by r e through resampling, and r 0m is shift data by r 0 through resampling; By introducing wedge matrices, equation (9) is represented by matrix notation, wherein only one pair of scaling parameters a i and b i in each two wedge matrices is non-zero, and the rest scaling parameters are zero, so that the number of non-zero values of the scaling parameters is equal to the number of wedge matrices, and the matrix notation is as follows: For each wedge matrix, the rearrangement matrix equation of equation (10) is converted into: The formula (11) is re-expressed as a matrix method: At this stage, U, V is interpreted as a circulant matrix; And carrying out wavelet convolution to obtain: And then rearranging: Wherein, the The length of a M ,b M is N+M+1 times N; the matrix symbols are then expressed as: Wherein, the Is of length N +. M+1 times nxm; M a 、m b has a length of N×M, and s has a length of N+M+1.
6. The non-linear constrained time domain high resolution spectral inversion method of claim 1 wherein said regularization constraint design objective function is constructed as follows: min(||s-Gm|| 2 +λ|m| 1 ) (16) Wherein s is input seismic data, G is a constructed model library, lambda is regularized parameters, and m is the result to be solved ,m＝[m a m b ] T ＝[a 0 a 1 …a M b 0 b 1 …b M ] T .
7. The non-linear constrained time domain high resolution spectral inversion method of claim 1 wherein solving the regularized non-linear problem using a fast iterative shrinkage threshold algorithm results in a high resolution reflection coefficient profile comprising: The nonlinear constraint time domain high-resolution spectrum inversion result is obtained through inversion, the iterative solution x k 、y k is initialized at first, the step length gamma 0 ＝1,t 0 =1 is achieved, and x k+1 is calculated in an iterative mode: Wherein, t k and a k are intermediate variables of iteration steps, and the objective is to provide each optimal iteration step to obtain the optimal iteration efficiency; y k ＝x k +a k (x k -x k-1 ) (18) Where sgn is a sign function, returns an integer variable indicating the sign of the parameter, max is the maximum value, and λ is the regularization parameter.
8. The non-linearly constrained time-domain high-resolution spectral inversion method of claim 7 wherein if ||x k+1 -x k | < epsilon then an iterative solution x k+1 is output, otherwise iterating (17) - (19) is repeated.
9. An electronic device, comprising a processor and a memory, wherein the processor is configured to execute a program of the nonlinear constrained time domain high resolution spectrum inversion method, so as to implement the spectrum inversion high resolution processing method based on strong axis stripping according to any one of claims 1 to 8.
10. A storage medium storing one or more programs executable by one or more processors to implement the non-linearly constrained time domain high resolution spectral inversion method of any one of claims 1-8.

Description

Nonlinear constraint time domain high-resolution spectrum inversion method Technical Field The invention relates to the field of seismic data broadband processing, in particular to a nonlinear constraint time domain high-resolution spectrum inversion method. Background Along with the gradual turning of oil and gas resource exploration targets in China to complex oil and gas reservoirs such as thin layers and unconventional oil and gas reservoirs, the conventional seismic data has low resolution due to the influence of wave group interference of thin layer structures and stratum energy absorption factors, and particularly in the aspect of thin layer identification, the conventional seismic data attribute extraction and interpretation technology has obviously insufficient resolution, so that certain difficulty is caused to thin layer seismic identification, and the fine requirements of exploration and development cannot be met. Aiming at the problem of low resolution of seismic data, the influence of seismic wavelets can be eliminated by utilizing a spectrum inversion technology, a high-resolution seismic wave reflection coefficient profile is obtained, and thin layer identification is realized. However, the conventional method generally has the defects of multi-resolution problem, lower resolution or larger calculation amount in the data processing process, and cannot meet the current industrial production requirements of high-precision reservoir identification. Therefore, how to develop a spectrum inversion technology with higher resolution and more finely describe the underground geological structure, so as to more accurately predict the thickness of underground reservoir and the spatial distribution condition thereof, improve the accuracy of stratum interpretation, and simultaneously maintain higher calculation efficiency, thus being one of the core problems for improving the resolution of seismic data. In China patent application No. CN201610020556.6, "an inversion method of seismic reflection coefficient based on total variation minimization constraint", a similar inversion method of reflection coefficient is mentioned, which is to invert post-stack seismic data by adopting a time window by time window. For a single time window, firstly, carrying out Fourier transformation on post-stack seismic data and seismic wavelets extracted in advance in the time window, then obtaining a frequency domain expression of a reflection coefficient, then carrying out Fourier transformation on the time-domain reflection coefficient, extracting a real part and an imaginary part of the time-domain reflection coefficient on the basis of parity decomposition of the reflection coefficient, constructing a corresponding solving equation, carrying out equation solving by adopting minimum total variation constraint on the basis of a traditional conjugate gradient algorithm, obtaining odd and even reflection coefficients, reconstructing the original reflection coefficient in the time window, and sequentially carrying out inversion on all the time windows to obtain the reflection coefficient of a single-channel seismic record. And carrying out inversion on all the seismic channels in sequence to obtain the reflection coefficient of each seismic record. The method can effectively invert the seismic reflection coefficient, and is beneficial to improving the resolution of seismic data and the prediction precision of a reservoir. However, in the application, a channel-by-channel time window inversion mode is adopted, so that frequency domain and time domain transformation is needed to be carried out on the seismic data in each time window, and a corresponding solving equation is constructed. Such a calculation is relatively large, resulting in a relatively long time-consuming inversion process, while the processing results shown in the present document indicate that the method has relatively poor lateral continuity. In the Chinese patent application No. CN201811031301.5, a reflection coefficient inversion method for improving the resolution of seismic data is mentioned, which is an inversion method based on L1 norm, and is used for inverting the seismic reflection coefficient, firstly, a low-frequency model is introduced on the basis to control the amplitude variation range of the reflection coefficient, then a smooth matrix is introduced, the variation of the seismic reflection coefficient before and after the introduction of a low-pass filter is compared, when the low-pass filter is constructed, firstly, the frequency spectrum is constructed in a frequency domain, then the frequency spectrum is transformed into a time domain, a filter function is obtained, finally, a filter matrix is constructed, and a linear equation after the introduction of the low-frequency model to control the reflection coefficient is solved, so that the reflection coefficient array in the formula can be obtained. The method of the invention improves the freque