Pre-stack gather optimization method and system based on multi-dimensional constraint dynamic time warping

Abstract

The invention belongs to the technical field of seismic exploration, and discloses a pre-stack gather optimization method based on multi-dimensional constraint dynamic time warping, which remarkably improves the space continuity and stability of a time shift field. According to the method, the time shift field which is smoother and more continuous in a three-dimensional space can be generated, and the common abnormal jump and unreasonable jitter of the time shift field in a conventional DTW method are effectively inhibited. This better conforms to the gradient characteristics of the geologic structure. Mismatching and processing illusion are effectively avoided, the adaptability of a complex structure area is improved, the global optimization capacity of the DTW algorithm is enhanced through multi-dimensional constraint, and the phenomenon that different horizons or effective waves are wrongly matched with interference waves due to local waveform complexity or noise interference is reduced. Therefore, common processing illusions such as event dislocation and local distortion in a conventional DTW method can be remarkably reduced.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of seismic exploration, and particularly relates to a pre-stack gather optimization method and system based on multi-dimensional constraint dynamic time warping. BACKGROUND

[0002] In seismic exploration, the quality of pre-stack CRP gathers is crucial for subsequent stacking imaging, migration imaging, velocity analysis, AVO analysis, and reservoir inversion. However, actual seismic data acquisition is often affected by various factors, such as wavefield propagation effects caused by the complexity of underground media, anisotropy, noise during acquisition, and inaccurate velocity models used for migration. These factors work together to cause problems such as residual moveout (incomplete flattening), waveform stretching and distortion in shallow layers for far-offset traces, and inconsistent phases in the same phase axis in CRP gathers, which seriously damages the consistency of the same phase axis in CRP gathers and affects the stacking effect and the accuracy of reservoir prediction.

[0003] For the problem of flattening the same phase axis in pre-stack gathers (gather optimization), existing processing methods can be roughly divided into two categories: (1) Correction method based on velocity model: This method optimizes the flattening effect of the gather by iteratively updating the velocity model (such as residual velocity analysis, tomographic inversion, etc.). However, iterative updating of the velocity model not only requires a large amount of computation and consumes a lot of computing resources and labor costs, but also often fails to completely eliminate the residual delay of the same phase axis on the gather in complex structural areas or anisotropic development areas. In addition, this method usually cannot effectively solve the waveform distortion problem of shallow far-offset seismic traces caused by dynamic calibration stretching and other factors.

[0004] (2) Data-driven method based on waveform similarity: This method does not rely on accurate velocity model updating, but directly calculates the local relative time shift between adjacent seismic traces and corrects the seismic traces to achieve alignment of the same phase axis. The key to this method is to design a stable and accurate seismic trace time shift estimation algorithm. Since it does not rely on velocity model updating, it usually has lower computational cost and higher processing efficiency.

[0005] In the gather optimization algorithm based on waveform similarity, the dynamic time warping (DTW) algorithm has received widespread attention because it can effectively handle the nonlinear alignment and matching problem of time series and to some extent, it balances waveform matching accuracy and computational efficiency.

[0006] (1) Conventional DTW algorithm and its application in the field of seismic exploration.

[0007] The DTW algorithm was originally applied in the field of speech recognition, and then introduced into seismic exploration. For example, it is used to calculate the correlation between strata; to achieve high-precision registration of PP wave and PS wave seismic profiles; to calibrate wells and seismic; to perform NMO correction to improve the stretching distortion of seismic traces at long offsets; and to be directly used for prestack trace flattening. In addition, the DTW is also used to extract fault information and automatically track horizons.

[0008] When the conventional DTW algorithm is applied to prestack trace (especially CRP trace) flattening, a single trace independent processing or a trace-by-trace serial processing mode is usually adopted: a reference trace (such as a near-offset trace or a stack trace) is selected from the CRP trace set, then the time shift sequence between the other traces and the reference trace (or the previous trace) is calculated one by one, and the time shift sequence is applied to correct the current trace to achieve the flattening of the event.

[0009] (2) Other waveform similarity methods related to DTW.

[0010] There are mainly three methods: one is the event flattening algorithm based on point-to-point mapping, which is simple to implement, but only calculates the overall time shift for single seismic data, which may lead to good matching for strong reflections and poor matching for weak reflections; two is the cross-correlation flattening algorithm under time window restriction, which depends on the selection of fixed time window, and is prone to event breakage or false matching; three is the absolute value cross-correlation trace flattening algorithm, which is relatively less used in practice.

[0011] (3) Improvement attempts aiming at the limitations of conventional DTW.

[0012] Further improvement in stability under low signal-to-noise ratio conditions is needed, for example, by applying recursive error accumulation (with smoothing effect) to the alignment error calculated for each sample point, and introducing time shift change rate constraint to improve the accuracy of estimated time shift; or designing a segmented DTW strategy to divide seismic traces into several sub-sections according to waveform similarity and amplitude strength, etc., to improve noise resistance by using the segmented time shift mean value combined with interpolation.

[0013] Despite the above applications and improvements, the conventional DTW and some improved methods still face the problems of local optimization, processing artifacts, spatial discontinuity, and noise sensitivity when dealing with complex actual seismic data, as described above. The multi-dimensional constraint DTW method proposed in the present application is aimed at the shortcomings of the prior art, and improves the stability and practicality of the DTW algorithm in trace set optimization by introducing regularization constraints in the spatial domain.

[0014] The quality of pre-stack common reflection point (CRP) gathers directly affects the accuracy of subsequent stack imaging and the reliability of reservoir inversion. However, actual seismic data is often affected by complex wavefield propagation effects, noise interference, migration velocity errors, and far-offset waveform stretching, etc., resulting in insufficient flattening of the events in the CRP gathers, poor phase consistency, and waveform distortion. These problems seriously affect the stacking quality and amplitude-preservation performance of the CRP gathers.

[0015] As a data-driven method based on waveform similarity, dynamic time warping (DTW) can effectively align and match two time series, and to some extent, can improve the consistency of the events in the CRP gathers. However, the conventional DTW method has the following main technical problems when applied to seismic gather optimization: Local optimization and processing artifacts: The conventional DTW algorithm usually adopts a single-channel independent or channel-by-channel serial processing mode, and only pursues the best alignment of local waveforms when matching adjacent seismic channels. This approach is easily disturbed by noise, and when the waveforms of adjacent channels do not strictly satisfy the stretching assumption of DTW in actual seismic data (for example, there are complex wavefield interference, migration arc, etc.), it is easy to incorrectly match waveforms of different layers or different properties, resulting in local distortion, event breakage or jumping, etc. processing artifacts, reducing the practical value of the method and the reliability of the processing results.

[0016] Lack of spatial continuity constraint leads to unstable time shift field: The conventional DTW method mainly considers the constraints in the time domain (such as the time shift rate) when calculating the time shift, and ignores the lateral continuity that the seismic signal should have in the spatial domain (such as the line and tie line directions). This leads to abnormal jumps and discontinuities in the spatially calculated time shift field, which destroys the lateral consistency of the events, especially in complex structural areas or low signal-to-noise ratio areas.

[0017] Sensitive to noise, performance decreases under low signal-to-noise ratio conditions: Random noise in seismic data can seriously interfere with the path search process of the DTW algorithm, resulting in local registration errors. In low signal-to-noise ratio areas, the conventional DTW algorithm may even force the matching of random noise, resulting in pseudo-coherence, further reducing the signal-to-noise ratio of the processed gathers and stack sections.

[0018] The main shortcomings of the prior art are: Spatial discontinuity caused by single-channel processing mode: The conventional DTW algorithm is usually processed in a single-channel mode, i.e. calculating the time shift between the current channel and the reference channel (or the previous channel). This processing mode only pursues the local optimal matching between adjacent channels, and lacks the global spatial continuity constraint on the time shift field of the entire channel set or adjacent channel set (such as on the common offset channel set COG). The reflection interface in the real underground usually has spatial continuity, and the corresponding time shift field should also be gradually changed in space. The time shift field generated by the conventional DTW may have a sharp jump in space, which destroys the lateral continuity of the event and causes false events on the stacked section.

[0019] Sensitivity to complex wave field and noise, easy to produce mismatch and processing artifacts.

[0020] The basic assumption of the DTW algorithm is that there is a strict (nonlinear) scaling relationship between the two sequences to be matched. However, due to the effects of complex wave field propagation (such as diffraction, multiple wave interference), offset arc, anisotropy and random noise, adjacent seismic channels in the actual CRP channel set do not strictly meet this assumption.

[0021] In this case, the path search mechanism of the conventional DTW is easy to incorrectly match the reflection waveforms of different geological horizons, or even the effective wave and the interference wave (such as noise), resulting in local overcorrection or undercorrection, and thus causing the breakage, distortion or false structure of the event on the processed channel set and stacked section.

[0022] In low signal-to-noise ratio areas, random noise will seriously interfere with the matching process of DTW, which may cause the forced matching of random noise waveforms and thus reduce the signal-to-noise ratio.

[0023] Local optimal problem of time shift estimation: The conventional DTW often falls into a local optimal solution when searching for the optimal matching path, and cannot guarantee the best matching in a global sense, especially in the case of complex and variable waveforms or the presence of multiple similar wave groups.

[0024] Error accumulation and propagation: In the single-channel serial processing mode, if the DTW matching of a channel has a large error, this error may be passed on and amplified in the processing of subsequent channels, affecting the optimization effect of the entire channel set.

[0025] Through the above analysis, the problems and defects of the prior art are: (1) Local optimal and processing artifacts.

[0026] (2) Lack of spatial continuity constraint leads to unstable time shift field.

[0027] (3) Sensitive to noise, performance decreases under low signal-to-noise ratio conditions. SUMMARY

[0028] In view of the problems of the prior art, the application provides a pre-stack gather optimization method based on multi-dimensional constraint dynamic time warping.

[0029] The application is implemented as follows: a pre-stack gather optimization method based on multi-dimensional constraint dynamic time warping comprises the following steps: Step 1: taking a common offset gather COG as a basic processing unit, introducing a regularization constraint in a three-dimensional time-space domain, and improving a DTW optimization target.

[0030] Step 2: estimating an initial two-dimensional or three-dimensional time shift field u for the current processing COG.

[0031] Step 3: smoothing and stabilizing the initial time shift field u by solving a regularization least square optimization problem to obtain a final optimized time shift field v.

[0032] Further, the DTW optimization target is improved.

[0033] The DTW method is a method for maximizing the similarity between two time series by locally stretching or compressing the two time series; considering two seismic records and , the difference between the two seismic records is a time shift sequence , that is, the following formula is satisfied: (1) That is, each sample of the seismic record is moved by samples to obtain the seismic record , and the time shift sequence characterizes a nonlinear scaling relationship; the target is to calculate the time shift sequence by the seismic records and ; in view of the spatial discontinuity and mismatching problems of the conventional DTW, the application takes a common offset gather (COG) as a basic processing unit and introduces a regularization constraint in a three-dimensional time-space domain; the improved optimization target of the application can be expressed as: (2) wherein is an integer in , represents the displacement of the two seismic records at the sample , and is an alignment error, representing the difference between the amplitudes of the two seismic records after the sample is displaced by samples, and it can be known from formula (1) that when ​This alignment error for ; at each sample point Calculation If there are several alignment errors, then all alignment errors will constitute The matrix, the real time-shifted sequence It is a path in the alignment error matrix, and all alignment error values ​​on this path are... ; in the constraints It is the first difference of the time-shifted sequence, and this condition guarantees the time-shifted sequence It cannot change drastically, among which The larger, The slower the change; These represent the sample points at... Indices in three directions; by solving this optimization problem, the time-shifted field that satisfies transverse continuity can be obtained. .

[0034] Furthermore, for the COG being processed, an initial two-dimensional or three-dimensional time-shifting field u is estimated.

[0035] However, the constrained optimization problem represented by Equation (2) is an NP-complete problem, which means that there is almost no computationally feasible solution. Therefore, this invention designs an efficient and stable approximation algorithm to adapt to the scale of actual seismic data. First, the initial time-shift field needs to be estimated based on the improved DTW algorithm.

[0036] For the currently processed COG (or COG pair, such as between the k-th COG and the (k-1)-th COG), calculate an initial time-shift field. This step can use the conventional DTW method. For example, within the COG, select a reference channel and then calculate the time shift of each other channel within the COG relative to the reference channel; or calculate the time shift field between the k-th COG and the (k-1)-th COG (the result of this calculation is the relative time shift between the two COGs). This stage mainly focuses on the matching of the time direction and can temporarily ignore the strong constraints of the spatial direction.

[0037] Therefore, ignoring the constraints in the xline and inline directions and retaining only the time constraints, equation (2) can be simplified to: (3) The essence of this optimization problem is: in a given length of... Width Alignment error matrix From arrive Find this continuous line where all error values ​​are 0. the path; considering that the actual problem usually contains noise, the objective is modified to find a continuous and cumulative error-minimized path, and the path should satisfy that the curvature change cannot be too drastic.

[0038] Further, the initial time shift field u is smoothed and stabilized.

[0039] After the initial time shift field is estimated by the improved DTW algorithm, the time shift field A new time shift field is obtained by applying three-dimensional regularization constraints , i.e., let the new time shift field be the closest to the original time shift field under the condition of satisfying the three-dimensional regularization constraints; this problem can be modeled as the following optimization problem: (4) In actual solving, the three-dimensional gradient constraint condition in formula (4) is converted into the gradient approaching 0 in three directions, which is equivalent to the parameters approaching infinity; by relaxing the strict constraint condition, the problem is converted into a convex optimization model that can be efficiently solved; therefore, the optimization problem of formula (4) can be rewritten as a regularization least square problem in vector form: (5) where is the first-order difference joint matrix in t, xline, and inline directions: (6) Taking the time direction as an example, the difference matrix of the single-channel time shift field can be constructed as: (7) The difference matrix of the entire COG gather time shift field in the time direction is the diagonalization expansion of the single-channel difference matrix : (8) Similarly, the difference matrices and in the xline and inline directions are constructed similarly to , but the position relationship between each sample and its adjacent samples in the matrix is more complex, which is not listed here; the Lagrange multiplier method is introduced to the optimization problem (5), which can be converted into solving a linear system: (9) where is the regularization parameter, which controls the constraint strength; since To meet the symmetry positive definiteness, the conjugate gradient algorithm can be used to iteratively solve the displacement field The method is applied to the current COG gather, and the correction of the COG gather is completed.

[0040] Another object of the present application is to provide a pre-stack gather optimization system based on multi-dimensional constraint dynamic time warping, comprising: An improvement module is used to take the common offset gather (COG) as a basic processing unit, introduce the regularization constraint in the three-dimensional time-space domain, and improve the DTW optimization target.

[0041] An estimation module is used to estimate an initial two-dimensional or three-dimensional time shift field u for the current processed COG.

[0042] A solving module is used to smooth and stabilize the initial time shift field u by solving a regularization least square optimization problem, and obtain a final optimized time shift field v.

[0043] Another object of the present application is to provide a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the computer program is executed by the processor to enable the processor to perform the steps of the pre-stack gather optimization method based on multi-dimensional constraint dynamic time warping.

[0044] Another object of the present application is to provide a computer readable storage medium storing a computer program, and the computer program is executed by a processor to enable the processor to perform the steps of the pre-stack gather optimization method based on multi-dimensional constraint dynamic time warping.

[0045] Another object of the present application is to provide an information data processing terminal for implementing the pre-stack gather optimization system based on multi-dimensional constraint dynamic time warping.

[0046] In combination with the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solutions to be protected by the present application are analyzed from the following aspects: Firstly, the present application aims to solve the above technical problems, and proposes an improved DTW algorithm based on multi-dimensional constraint. By constructing a time-space domain joint constraint model on the common offset gather (COG), regularization constraints are imposed on the calculated time shift field in the time direction, the contact survey line direction and the main survey line direction, so as to eliminate the abnormal changes of the time shift field, enhance the consistency and stability of the time shift field in the global sense, effectively avoid the mismatch phenomenon of the conventional DTW, improve the in-phase property of the pre-stack gather and the continuity of the stacking profile, and especially show stronger robustness in complex structure areas and low signal-to-noise ratio environments.

[0047] In view of the defects of the prior art, the purpose of the present application is to provide a more stable and robust dynamic time warping gather optimization method, which specifically comprises: 1. Enhance the spatial continuity and stability of the time shift field: by taking the common offset gather (COG) as the processing unit and introducing multi-dimensional regularization constraints in the time direction, the main line direction and the liaison line direction, the abnormal jump of the time shift field is suppressed, and the calculated time shift field has good smoothness and continuity in three-dimensional space, so that the lateral consistency of the seismic signal is better maintained.

[0048] 2. Improve the processing effect under complex geological conditions and low signal-to-noise ratio environment: by multi-dimensional constraint, the sensitivity of DTW algorithm to noise and complex waveform is reduced, and the error matching of effective wave and interference wave or waveform of different layers is avoided, so that the generation of processing artifacts (such as same-phase axis fault, local distortion, etc.) is effectively reduced.

[0049] 3. Improve the phase consistency of prestack gather and the quality of stack profile: by more accurate and stable time shift correction, the flattening effect and phase consistency of the same phase axis in prestack gather are significantly improved, and the imaging accuracy, signal-to-noise ratio and continuity of the same phase axis of the stack profile are improved, providing a higher quality data basis for subsequent seismic interpretation and reservoir inversion.

[0050] 4. Provide a new way for prestack gather optimization in complex structure area: the proposed method aims to overcome the limitations of conventional DTW in complex structure and low signal-to-noise ratio area, and provides a more reliable and effective gather optimization technology for seismic data processing in these areas.

[0051] Secondly, as the creative auxiliary evidence of the claims of the present application, it is also embodied in the following important aspects: The technical scheme of the present application fills the technical gap in the field of seismic data processing, which improves the dynamic time warping (DTW) algorithm from local one-dimensional waveform matching to global three-dimensional spatial collaborative optimization.

[0052] In geophysical exploration, dynamic time warping (DTW) is a powerful waveform matching tool that is very suitable for correcting the residual time difference of prestack gather in theory. However, a long-standing technical bottleneck is that the conventional DTW method is essentially a local optimization tool that "fights alone": it only processes single or adjacent seismic traces in isolation, completely ignoring the spatial continuity that seismic reflection events should have in geology. This mode leads to discontinuous and unstable results in space, which severely limits its application effect and reliability in actual production. Therefore, there has been a lack of mature technical solutions that can make the DTW algorithm play its waveform matching advantage while following the geological rules and conducting global spatial constraints.

[0053] The present application first proposes a multi-dimensional constraint DTW optimization framework (as claimed in claims 1-4), the core of which is to promote the processing unit from a single channel to a common offset gather (COG), and creatively introduces three-dimensional joint regularization constraints in the time, main survey line and contact survey line directions. This makes the calculation of the time shift field no longer a collection of isolated points, but a solving process of a three-dimensional field that should satisfy spatial smooth continuity as a whole. This effectively fills the gap in the industry of transforming local DTW algorithms into global, spatially continuous, geologically meaningful channel set optimization techniques, and opens up a new, more reliable and practical development direction for data-driven channel set optimization methods.

[0054] The technical solution of the present application solves the technical problems of "processing artifacts" and "spatial discontinuity" that have existed for a long time and have not been effectively solved when applying dynamic time warping (DTW) for pre-stack channel set optimization.

[0055] In the field of seismic data processing, researchers have always been eager to find a data-driven method that can automatically, accurately and stably flatten the pre-stack channel event. Although the DTW algorithm has potential, in actual application, its "single channel processing" mode makes it extremely sensitive to noise and complex wave fields, and it is easy to fall into a local optimal solution, resulting in the incorrect matching of reflection waves from different geological horizons, even effective waves and noise. This will generate a large number of processing artifacts, such as abnormal jumps, breaks and local distortions of events, and the processed results may even be more difficult to interpret than the original data, which has become the core problem hindering the widespread and reliable application of DTW technology in the industry.

[0056] The present application successfully solves this problem through its innovative "estimation first, optimization second" two-step strategy. In the first step, the conventional DTW method is allowed to generate an initial time shift field containing noise and discontinuous jumps; in the second step (core step), the initial time shift field is "reshaped" and "smoothed" by solving a global, multi-dimensional constraint regularization least squares problem (as shown in formulas 5-9). This process can effectively identify and suppress abnormal time shift values that are not continuous in space and do not conform to the geological gradual change rule, while preserving the overall time shift trend. As shown in the comparison of Figure 4 and Figure 7 , the time shift field generated by the conventional DTW has a sharp jump, while the time shift field generated by the present application is smooth and continuous. This stabilization of the time shift field directly leads to the elimination of processing artifacts in Figure 6 (c) part, and finally obtains the geologically reasonable and continuous event imaging result shown in Figure 6 (b) part. Therefore, the present application successfully solves the industry-recognized technical problem of spatial discontinuity artifacts caused by local mismatching of the DTW algorithm. BRIEF DESCRIPTION OF DRAWINGS

[0057] Figure 1 is a pre-stack gather optimization method flow chart based on multi-dimension constraint dynamic time warping provided by the embodiment of the application.

[0058] Figure 2 is a pre-stack gather optimization system structure block diagram based on multi-dimension constraint dynamic time warping provided by the embodiment of the application.

[0059] Figure 3 is a test of the conventional DTW algorithm on work area A provided by the embodiment of the application; (a) part is the original CRP gather; (b) part is the CRP gather processed by the conventional single-channel mode DTW; (c) part is the original stack profile; (d) part is the stack profile processed by the conventional single-channel mode DTW.

[0060] Figure 4 is a test result of the improved DTW algorithm on work area A provided by the embodiment of the application; (a) part is the CRP gather processed by the improved DTW algorithm; (b) part is the stack profile processed by the improved DTW algorithm. Figure 3 is a time shift field (upper part of the figure) at offset = 1725 m corresponding to the stack profile in the middle of the figure and the change of the time shift field with Xline at 3.2 s (lower part of the figure).

[0061] Figure 5 is a test result of the improved DTW algorithm on work area A provided by the embodiment of the application; (a) part is the CRP gather processed by the improved DTW algorithm; (b) part is the stack profile processed by the improved DTW algorithm.

[0062] Figure 6 is a local magnification comparison of the stack profile of work area A provided by the embodiment of the application; (a) part is the original stack profile; (b) part is the stack profile processed by the improved DTW algorithm; (c) part is the stack profile processed by the conventional DTW algorithm.

[0063] Figure 7 is a comparison of a certain CRP gather of work area B provided by the embodiment of the application; (a) part is the original CRP gather; (b) part is the CRP gather processed by the improved DTW algorithm; (c) part is the CRP gather processed by the conventional DTW algorithm. Figure 4 is a time shift field (upper part of the figure) of the improved algorithm corresponding to the stack profile in the middle of the figure and the change of the time shift field with Xline at 3.2 s (lower part of the figure).

[0064] Figure 8 is a comparison of a certain CRP gather of work area B provided by the embodiment of the application; (a) part is the original CRP gather; (b) part is the CRP gather processed by the improved DTW algorithm; (c) part is the CRP gather processed by the conventional DTW algorithm.

[0065] Figure 9 is a comparison of the stack profile of work area B provided by the embodiment of the application; (a) part is the original stack profile; (b) part is the stack profile processed by the improved DTW algorithm; (c) part is the stack profile processed by the conventional DTW algorithm.

[0066] Figure 10 Figure 2 is a comparison of CRP gathers of a certain work area B provided by an embodiment of the present application; (a) is partially the original CRP gather; (b) is the CRP gather processed by the improved DTW algorithm; (c) is partially the CRP gather processed by the conventional DTW algorithm.

[0067] Figure 11 Figure 3 is a comparison of stacked sections of the work area B provided by an embodiment of the present application; (a) is partially the original stacked section; (b) is the stacked section processed by the improved DTW algorithm; (c) is partially the stacked section processed by the conventional DTW algorithm.

[0068] Figure 12 Figure 4 is a comparison of time shift fields calculated by the improved DTW algorithm and the conventional DTW algorithm provided by an embodiment of the present application; (a) is the time shift field profile of the COG gather at offset = 2075m (the upper part of the figure) and the change of time shift with Xline at 3.0s (the lower part of the figure) calculated by the improved DTW algorithm; (b) is the time shift field profile of the COG gather at offset = 2075m (the upper part of the figure) and the change of time shift with Xline at 3.0s (the lower part of the figure) calculated by the conventional DTW algorithm. DETAILED DESCRIPTION

[0069] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and should not be used to limit the present application.

[0070] *CRP gather (Common Reflection Point gather): a basic data organization form in seismic data processing, which collects seismic traces reflected from the same point underground.

[0071] *COG gather (Common Offset Gather): a seismic data organization form, which collects seismic traces with the same offset.

[0072] *DTW (Dynamic Time Warping): a method for measuring the similarity between two time series, especially the time series that may change at different speeds. It calculates the optimal match by "warping" the time axis.

[0073] *Time shift: the relative time difference between seismic traces or seismic events in different traces.

[0074] * TimeShiftField: A collection of time shift values calculated for each point or trace within a certain spatial extent (e.g. a gather or a profile), usually a 2D or 3D function.

[0075] * Regularization: In mathematics and statistics, a process of introducing additional information (constraints) to solve ill-posed problems or prevent overfitting. In this invention, it is used to constrain the time shift field to be smoother and more continuous.

[0076] * Seismic Event / Reflection: A continuous arrangement of traces on a seismic profile or gather that represents the reflection of a single geological interface or discontinuity in the subsurface, with similar waveform characteristics.

[0077] * Gather Optimization: Processing of pre-stack gathers to improve the flattening, waveform consistency and signal-to-noise ratio of events, thus improving the quality of subsequent stack imaging and reservoir inversion.

[0078] The present invention proposes a dynamic time warping (DTW) pre-stack gather optimization method based on multi-dimensional constraints. The core idea is to take common offset gather (COG) as the processing unit, and introduce regularization constraints in time (t), cross-line (xline) and inline directions during the calculation of time shift between seismic traces, to enhance the spatial continuity and stability of the time shift field, thus overcoming the limitations of conventional DTW algorithm.

[0079] As shown in Figure 1 , the pre-stack gather optimization method based on multi-dimensional constraint dynamic time warping provided by the embodiment of the present invention includes the following steps: S101, taking common offset gather COG as the basic processing unit, introducing regularization constraints in three-dimensional space-time domain, and improving the DTW optimization target.

[0080] S102, for the current processed COG, estimating an initial two-dimensional or three-dimensional time shift field u.

[0081] S103, by solving a regularization least squares optimization problem, smoothing and stabilizing the initial time shift field u, to obtain the final optimized time shift field v.

[0082] The embodiment of the present invention improves the DTW optimization target.

[0083] The DTW method is a method of maximizing the similarity between two time series by locally stretching or compressing them; considering two seismic records and , , the difference between the two is a time shift sequence , that is, satisfies the following formula: (1) , that is, the seismic record of each sample point moves sample points to obtain the seismic record , the time shift sequence characterizes a nonlinear scaling relationship; the target is to calculate the time shift sequence and through the seismic record ; in view of the existing spatial discontinuity and mismatch problem of the conventional DTW, the present application proposes to take the common offset gather (COG) as a basic processing unit, and introduces the regularization constraint of three-dimensional space-time domain; the improved optimization target of the present application can be expressed as: (2) , wherein is an integer in , representing the displacement of two seismic records at sample point , and is an alignment error, representing the difference between the amplitudes of two seismic records after moving sample points at sample point , as can be seen from formula (1), when , the alignment error is ; at each sample point , the alignment error is calculated, so that all alignment errors will form a matrix of , and the real time shift sequence is a path in the alignment error matrix, and all alignment error values on the path are ; in the constraint condition, is the first-order difference of the time shift sequence, which ensures that the time shift sequence cannot change drastically, wherein is larger, the change is slower; respectively represent the indices of the sample points in the three directions; by solving the optimization problem, the time shift field satisfying the lateral continuity can be obtained. .

[0084] The embodiment of the present application estimates an initial two-dimensional or three-dimensional time shift field u for the current processed COG.

[0085] However, the constraint optimization problem represented by formula (2) belongs to NP-complete problem, which means that there is almost no computationally feasible solution; therefore, the present application designs a high-efficiency and stable approximate algorithm to adapt to the scale of actual seismic data, and first needs to estimate an initial time shift field based on the improved DTW algorithm.

[0086] For the current processing COG (or COG pair, such as the kth COG and the k-1th COG), an initial time shift field is calculated ; this step can adopt a conventional DTW method, for example, selecting a reference trace in the COG, and then calculating the time shift of other traces in the COG relative to the reference trace; or calculating the time shift field between the kth COG and the k-1th COG (at this time, the calculation result is the relative time shift between the two COGs); this stage mainly focuses on the matching in the time direction, and can temporarily ignore the strong constraint in the space direction.

[0087] Therefore, after ignoring the constraints in the xline and inline directions and only retaining the constraint in the time direction, formula (2) can be simplified as: (3) The essence of the optimization problem is: in the alignment error matrix with a length of and a width of , find a continuous path from to with an error value of ; considering that the actual problem usually contains noise, therefore, when constructing the optimization problem, the target is modified to find a continuous path with the minimum accumulated error, and the path needs to satisfy that the change in bending cannot be too drastic.

[0088] The present application embodiment provides smoothing and stabilizing processing on the initial time shift field u.

[0089] After the initial time shift field is estimated by the improved DTW algorithm, the time shift field is required to be applied with three-dimensional regularization constraints to obtain a new time shift field , that is, the new time shift field is closest to the original time shift field under the condition of satisfying the three-dimensional regularization constraints; this problem can be modeled as the following optimization problem: (4) In actual solving, the three-dimensional gradient constraint condition in formula (4) is converted into the gradient in three directions tending to 0, which is equivalent to The parameter approaches infinity; by relaxing the strict constraint condition, the problem is converted into a convex optimization model which can be solved efficiently; therefore the optimization problem of formula (4) can be rewritten as a vector form of regularized least square problem: (5) Where is the first-order difference joint matrix in t, xline, inline directions: (6) Taking the time direction as an example, the difference matrix of single-channel time-lapse field can be constructed as: (7) The difference matrix of the entire COG gather time-lapse field in the time direction is the diagonalization expansion of the single-channel difference matrix : (8) Similarly, the construction of the difference matrices and in the xline and inline directions are similar to , but the position relationship between each sample and its adjacent samples in the matrix is more complex, which is not listed here; the introduction of Lagrange multiplier method to the optimization problem (5) can be converted into solving a linear system: (9) Where is the regularization parameter, which controls the constraint strength; since satisfies the symmetry positive definite property, the conjugate gradient algorithm can be used to solve the displacement field iteratively, and the displacement field is applied to the current COG gather to complete the correction of the COG gather.

[0090] Construction method of joint difference operator for time-lapse field regularization The operator combines the first-order (or higher-order) difference information in the time direction, the main survey line direction and the liaison survey line direction, and is used to impose spatial and temporal smoothing constraints in the optimization problem.

[0091] The calculation scheme of the optimized time-lapse field is obtained by solving the linear equation (formula 9), and the optimized time-lapse field with spatial continuity is applied to the corresponding COG gather to correct the seismic trace, improve the flattening degree of the event and the waveform consistency.

[0092] As Figure 2As shown, the embodiment of the present application provides a pre-stack gather optimization system based on multi-dimensional constraint dynamic time warping, which comprises: An improvement module is used to take the common offset gather COG as a basic processing unit, introduce the regularization constraint of three-dimensional time-space domain, and improve the DTW optimization target.

[0093] An estimation module is used to estimate an initial two-dimensional or three-dimensional time shift field u for the current processing COG.

[0094] A solving module is used to smooth and stabilize the initial time shift field u by solving a regularization least square optimization problem, and obtain the final optimized time shift field v.

[0095] Another object of the present application is to provide a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the computer program is executed by the processor to make the processor execute the steps of the pre-stack gather optimization method based on multi-dimensional constraint dynamic time warping.

[0096] Another object of the present application is to provide a computer readable storage medium storing a computer program, and the computer program is executed by the processor to make the processor execute the steps of the pre-stack gather optimization method based on multi-dimensional constraint dynamic time warping.

[0097] Another object of the present application is to provide an information data processing terminal for realizing the pre-stack gather optimization system based on multi-dimensional constraint dynamic time warping.

[0098] The present application is specifically implemented: Compared with the existing DTW gather optimization method mainly adopting the conventional single channel or channel-by-channel serial processing mode, the improved DTW method based on multi-dimensional constraint proposed by the present application has the following obvious advantages: 1. Significantly improve the spatial continuity and stability of the time shift field: by taking the COG as the processing unit and introducing the joint regularization constraint of time, main survey line and contact survey line direction, the present application can generate a time shift field which is more smooth and continuous in three-dimensional space (as shown in (a) part of Figure 7 , Figure 12 (b) part of Figure 4 , Figure 12 ), effectively inhibiting the abnormal jump and unreasonable jitter of the time shift field commonly seen in the conventional DTW method (as shown in (b) part of Figure 4 , Figure 12 ). This is more in line with the gradual change characteristics of the geological structure.

[0099] 2. Effectively avoid false matching and processing artifacts, improve adaptability in complex structure area: Multi-dimensional constraints enhance the global optimization ability of DTW algorithm, reducing the phenomenon of false matching of different layers or effective waves and interference waves caused by local waveform complexity or noise interference. Therefore, the invention can significantly reduce the processing artifacts such as common event break and local distortion in conventional DTW method Figure 6 Part (b) of (c) in Figure 6 Part (b) of (c) in Figure 9 Part (b) of (c) in Figure 11 Part (b) of (c) in Figure 11 Part (b) of (c) in

[0100] 3. Enhance the robustness to noise and improve the processing effect of low signal-to-noise ratio data: Spatial constraints make time shift estimation no longer rely on single channel or adjacent two channels, but consider the information of surrounding channels, thereby improving the resistance to random noise. The invention can avoid the problem of forced matching of random noise in conventional DTW under strong noise background, thereby generating pseudo-coherence (as shown in Figure 10 Part (c) of (c) in Figure 11 Part (c) of (c) in

[0101] 4. Overall improve the quality of pre-stack gather and stack profile: Due to the improvement of accuracy and stability of time shift correction, the pre-stack gather processed by the invention has better event flattening effect and phase consistency (as shown in Figure 5 Part (a) of (b) in Figure 8 Part (b) of (b) in Figure 10 Part (b) of (b) in Figure 5 Part (b) of (b) in Figure 6 Part (b) of (b) in Figure 9 Part (b) of (b) in Figure 11 Part (b) of (b) in

[0102] 5. Provide a more reliable data basis for subsequent processing: High-quality pre-stack gather and stack profile are the key to subsequent seismic processing and interpretation work such as precise velocity analysis, AVO inversion and reservoir prediction. The invention provides a solid data basis for these subsequent links by providing better gather optimization results.

[0103] 6. Strong practicability, providing a new effective way for gather optimization under complex geological conditions: By introducing regularization and convex optimization strategy, the three-dimensional constrained optimization problem which is difficult to solve directly is transformed into a linear problem which can be solved efficiently, so that it can be applied to the processing of actual large-scale seismic data, especially in complex structures and low signal-to-noise ratio areas, showing strong robustness and application value.

[0104] The application effectively overcomes the main limitations of the conventional DTW method in prestack gather optimization by introducing multi-dimensional space-time constraints into the DTW algorithm, and can significantly improve the quality and reliability of the processing results.

[0105] The multi-dimensional constraint dynamic time warping based prestack gather optimization method proposed in the application can be widely applied in the following fields or related products: Oil and gas exploration data processing: As a core application field, the application can be used in the prestack seismic data processing flow in conventional and unconventional oil and gas exploration, and the CRP gather is optimized before stacking, migration and velocity analysis, which significantly improves the signal-to-noise ratio and resolution of the final imaging profile.

[0106] Reservoir prediction and AVO analysis: By providing prestack gathers with better flattening effect and better amplitude preservation, the application provides a high-quality data basis for subsequent amplitude versus offset (AVO) analysis, elastic parameter inversion and fluid identification for reservoir characterization.

[0107] Mineral resources and geothermal exploration: It can be applied to the seismic exploration of metal mines, coal and geothermal resources, and through improving the gather quality, clearer ore-controlling structure and thermal reservoir imaging can be obtained.

[0108] Engineering investigation and environmental monitoring: In the fields of urban shallow seismic exploration, major engineering site selection and active fault detection, the method can be used to improve the imaging accuracy of shallow reflection.

[0109] Four-dimensional (4D) time-lapse seismic monitoring: In the time-lapse seismic monitoring of oil and gas field development, the method can be used to correct the time difference between seismic data collected at different times with high precision, and to support reservoir dynamic change monitoring.

[0110] Related products: The method of the application can be integrated into various commercial or academic geophysical data processing software systems as a core functional module, providing a more advanced and reliable prestack gather optimization tool for seismic data processing personnel.

[0111] The application processes actual industrial seismic data of two different work areas (work area A and work area B), and conducts a systematic and comprehensive comparison with the conventional DTW method, obtaining sufficient and direct technical effect evidence, proving the significant technical progress of the application.

[0112] 1. It fundamentally improves the spatial continuity of the time-shift field.

[0113] like Figure 12 As shown, the time-shift field profile calculated by the conventional DTW method ( Figure 12 (b) and slices along the Xline Figure 12 Part (b) below exhibits violent, irregular jittering and jumps. The time-shift field calculated by the method of this invention (…) Figure 12 Part (a) exhibits excellent smoothness and spatial continuity, which better reflects the gradual physical laws governing the changes in real underground geological bodies. This stabilization of the time-shift field is the fundamental reason for the excellent application results achieved by this invention.

[0114] 2. Significantly improved the quality of the overlay imaging profile and the clarity of the geological structure.

[0115] In the comparison of superimposed cross-sections of work area A and work area B ( Figure 6 , Figure 9 , Figure 11 The difference in results is quite significant. While the profile processed by conventional DTW (part (c) in the figure) shows some improvement compared to the original profile (part (a) in the figure), it still exhibits numerous instances of phase axis discontinuity and local distortion. In contrast, the superimposed profile processed by the method of this invention (part (b) in the figure) demonstrates excellent continuity of the stratigraphic phase axes, clear discontinuities, a more realistic and reliable structural morphology, and a significantly improved signal-to-noise ratio, resulting in a qualitative leap in overall imaging quality.

[0116] 3. Effectively eliminates the processing artifacts produced by conventional DTW.

[0117] In a comparison of the direct processing effects of pre-stack gathers ( Figure 8 , Figure 10 Conventional DTW methods (part (c) in the figure) introduce new local distortions and discontinuities into the gather due to mismatches when correcting the in-phase axis. However, the method of this invention (part (b) in the figure), with its spatial constraint capability, can perform smooth and continuous correction of the in-phase axis, effectively avoiding the generation of artifacts. The processed gather (e.g.) Figure 8 (b) In the middle part, the phase axis is perfectly flattened.

[0118] The processing results from multiple actual work areas irrefutably demonstrate, from multiple dimensions such as the physical characteristics of the time-shifted field, the flattening effect of the pre-stack gather, and the imaging quality of the final stacked profile, that the present invention has achieved decisive success in solving the inherent defects of conventional DTW algorithms in terms of spatial discontinuity and susceptibility to artifacts. Compared with existing technologies, it has huge advantages and significant beneficial effects.

[0119] It should be noted that embodiments of the present application can be realized by hardware, software, or a combination of software and hardware. The hardware portion can be realized by a special logic; the software portion can be stored in a memory and executed by a proper instruction execution system, such as a microprocessor or a specially designed hardware. A person of ordinary skill in the art can understand that the above-mentioned apparatus and method can be realized by computer executable instructions and / or included in processor control codes, such as a carrier medium, such as a magnetic disk, CD or DVD-ROM, a programmable memory, such as a read-only memory (firmware), or a data carrier, such as an optical or electronic signal carrier. The apparatus of the present application and its modules can be realized by a hardware circuit, such as a very large scale integrated circuit or a gate array, a semiconductor, such as a logic chip, a transistor, or a programmable hardware device, such as a field programmable gate array, a programmable logic device, or the like, by software executed by various types of processors, or by a combination of the above-mentioned hardware circuit and software, such as firmware.

[0120] The above description is merely a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any modification, equivalent replacement, and improvement within the technical range disclosed by the present application, and within the spirit and principle of the present application, should be covered within the protection scope of the present application.

Claims

1. A pre-stack gather optimization method based on multi-dimension constraint dynamic time warping, characterized in that, The method comprises the following steps: a. selecting a common offset gather (COG) as a basic processing unit, and adding a three-dimensional space regularization constraint to a dynamic time warping optimization target; b. estimating an initial two-dimensional or three-dimensional time shift field u for the current COG; c. establishing a regularization least square model with a three-dimensional gradient constraint, and performing smoothing and stabilization processing on the initial time shift field to obtain an optimized time shift field v; d. applying the optimized time shift field to the corresponding COG to complete gather alignment.

2. The method of claim 1, wherein, The regularization constraint adopts a first-order difference form, and applies smoothing limits to the gradients in the time direction, the inline direction, and the xline direction, to ensure that the time shift field is continuous and monotonic in the three-dimensional space.

3. The method of claim 1, wherein, The initial time shift field is obtained by retaining only the alignment error term and ignoring the gradient constraint term in the improved dynamic time warping target, and is used as an initial value for subsequent regularization least square solving.

4. The method of claim 1, wherein, The regularization least square model is converted into a linear equation set by a Lagrange multiplier method, the linear equation set has a symmetric positive definite property, and a conjugate gradient algorithm is used for iterative solving.

5. A system for pre-stack gather optimization based on multi-dimension constrained dynamic time warping, characterized in that, The method comprises: an improvement module for adding a three-dimensional space regularization constraint to a dynamic time warping optimization target; an estimation module for estimating an initial two-dimensional or three-dimensional time shift field for a current common offset gather; a solving module for constructing a three-dimensional gradient constraint regularization least square model and iteratively solving an optimized time shift field; an output module for applying the optimized time shift field to the common offset gather to realize gather alignment.

6. The system of claim 5, wherein, The improvement module generates a target path by constructing a cost function including an alignment error matrix and a first-order difference constraint, to realize lateral continuous time shift field searching.

7. A computer readable storage medium characterized in that, A computer program is stored thereon, and the program, when executed by a processor, executes the method of any one of claims 1 to 4.

8. A computing device, comprising: A processor and a memory are included, and a computer program is stored in the memory, and the computer program, when executed by the processor, causes the processor to execute the method of any one of claims 1 to 4.

9. An information data processing terminal, characterized by The information data processing terminal integrates the system of claim 5, and is used for performing multi-dimensional constraint dynamic time warping processing on seismic prestack data.

10. A seismic data processing platform, characterized by, The platform comprises a plurality of parallel computing nodes, the computing nodes deploy the computer program of claim 7, and perform the prestack gather optimization method based on multi-dimensional constraint dynamic time warping on a plurality of common offset gathers simultaneously in a distributed manner.

Images