Dynamic programming-based strike-slip fracture automatic tracking method and related equipment

Abstract

The invention discloses a strike-slip fracture automatic tracking method based on dynamic programming and related equipment, and the method comprises the steps: carrying out the signal-to-noise ratio estimation of seismic data, and obtaining the signal-to-noise ratio attribute of a strike-slip fault; calculating a probability density function of seismic data by taking a signal-to-noise ratio attribute as a constraint so as to obtain a maximum likelihood attribute of a strike-slip fault; screening an optimal path for the maximum likelihood attribute by using a dynamic programming algorithm, and finally carrying out geological interpretation on the obtained fracture identification attribute of the strike-slip fault, depicting the occurrence of the strike-slip fracture, and obtaining the fault position, the fault trend and the fault inclination angle of the strike-slip fault; compared with a traditional method, the method has the advantages that the fault attribute extraction mode is higher in certainty, the depiction of fractures in a section and a plane is more detailed and accurate, and meanwhile, the method has a good filtering effect on event disappearance or dislocation caused by non-faults; the method can clearly represent the vertical distribution of the strike-slip fault and determine the longitudinal development interval of the strike-slip fault, and has high reliability.

Description

Technical Field

[0001] This invention belongs to the field of geophysical technology for oil and gas exploration, specifically relating to an automatic tracking method and related equipment for strike-slip fractures based on dynamic programming. Background Technology

[0002] A strike-slip fault is a fracture that occurs within a strike-slip fault. A strike-slip fault is a fracture that occurs when the two sides of a fault move horizontally relative to each other under the action of a couple, under the influence of torsional or shear stress fields. The mechanical properties and structural patterns of strike-slip faults can change significantly in different evolutionary stages or different sections, and various strike-slip transformation structures can be formed; correspondingly, the composition, properties, and distribution characteristics of their fluids also vary. Accurate identification and interpretation of strike-slip faults are crucial for understanding the distribution, formation, and enrichment patterns of oil and gas reservoirs. Because carbonate oil and gas reservoirs are characterized by diverse reservoir types, large variations in reservoir space, and strong heterogeneity, their development presents many challenges, such as difficulties in reservoir description and fracture simulation, and challenges in reservoir numerical simulation. Therefore, accurate identification and interpretation of strike-slip faults is one of the key and challenging aspects of carbonate oil and gas reservoir development.

[0003] Currently, in the identification of strike-slip faults, while the plane resolution of fault-sensitive attributes is high and the fault distribution is relatively clear on the plane, the fault surface is difficult to directly identify in the seismic data volume due to the small longitudinal displacement of strike-slip faults. At the same time, the development intensity of faults is currently difficult to represent in the data volume; methods such as coherence enhancement, ant volume analysis, and likelihood analysis often fail to distinguish the fault order in the identified fault surfaces, and are influenced by many non-fault factors, resulting in a relatively messy fault characterization.

[0004] Therefore, current strike-slip fault identification technology cannot directly identify the fault surface in seismic data volumes, and it is difficult to distinguish the fault order of the fault surface. Summary of the Invention

[0005] To overcome the shortcomings of the above-mentioned technologies, this invention provides an automatic strike-slip fault tracking method and related equipment based on dynamic programming. This method can solve the technical problem that current strike-slip fault identification technologies are difficult to directly identify the fault surface in seismic data volumes and distinguish the fault order of the fault surface.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] An automatic strike-slip fracture tracking method based on dynamic programming includes:

[0008] Obtain the signal-to-noise ratio of strike-slip faults based on seismic data;

[0009] Using the signal-to-noise ratio attribute of the strike-slip fault as a constraint, the probability density function of the seismic data is calculated, and the maximum likelihood attribute of the strike-slip fault is obtained based on the calculation result of the probability density function.

[0010] The optimal path is selected for the maximum likelihood attribute using a dynamic programming algorithm, and the fracture identification attribute of the strike-slip fault is obtained based on the selected optimal path.

[0011] The geological interpretation of the fracture identification attributes is used to characterize the occurrence of the strike-slip fault and obtain the fault location, strike, and dip angle of the strike-slip fault.

[0012] Furthermore, before obtaining the signal-to-noise ratio attribute of the strike-slip fault based on the seismic data, the original seismic data is subjected to fault enhancement preprocessing to obtain the seismic data.

[0013] Furthermore, the specific steps for fault enhancement preprocessing of the original seismic data include:

[0014] The original seismic data were subjected to two-dimensional filtering using a preset denoising method to remove oblique interference noise.

[0015] In three-dimensional space, the covariance or variance of the original seismic data after two-dimensional filtering is processed to obtain the structural tensor corresponding to the original seismic data.

[0016] Perform eigenvalue decomposition on the structure tensor to obtain the structure tensor eigenvalues;

[0017] The diffusion tensor is calculated based on the eigenvalues ​​of the structural tensor, and the original seismic data is then subjected to diffusion filtering using the diffusion tensor as a constraint to obtain the seismic data.

[0018] Furthermore, the specific steps for performing covariance or variance processing on the original seismic data after two-dimensional filtering in three-dimensional space to obtain the structural tensor corresponding to the original seismic data include:

[0019] The raw seismic data is represented as a three-dimensional dataset F(x,y,z), where x,y,z represent spatial coordinates.

[0020] The original seismic data is cut into windows according to the preset cutting window size, and the covariance or variance of the original seismic data in each window is calculated to obtain the covariance matrix or variance matrix.

[0021] The covariance matrix within each window is averaged or accumulated to obtain the overall structural tensor of the original seismic data.

[0022] Furthermore, the specific steps for performing eigenvalue decomposition on the structure tensor to obtain the structure tensor eigenvalues ​​include:

[0023] According to the eigenvalue decomposition formula, the structural tensor is decomposed into three eigenvalues; the eigenvalues ​​represent the magnitude of the gradient changes in the original seismic data in the spatial and temporal directions.

[0024] Sort the three eigenvalues ​​in descending order, and then represent them as the first eigenvalue, the second eigenvalue, and the third eigenvalue based on the sorting result.

[0025] The second eigenvalue is used as the eigenvalue of the structure tensor, and the corresponding eigenvector is calculated.

[0026] Furthermore, the preset denoising method includes FK filtering, wavelet transform filtering, principal component analysis, and two-dimensional low-pass filtering.

[0027] Furthermore, the specific steps for calculating the probability density function of the seismic data using the signal-to-noise ratio attribute of the strike-slip fault as a constraint, and obtaining the maximum likelihood attribute of the strike-slip fault based on the calculation result of the probability density function, include:

[0028] The data points in the seismic data are represented as x. i , where i represents the index of the data point;

[0029] The signal-to-noise ratio of the strike-slip fault is used as a constraint condition for the Gaussian model, and the probability density function of the seismic data is calculated using the Gaussian model. The probability density function is expressed as follows:

[0030]

[0031] Wherein, P(x i |Fault) represents data point x i The probability of belonging to a fault, P(x) i (Non-fault) represents data point x i The probability of belonging to a non-fault region, u1 is the mean of the fault region, u2 is the mean of the non-fault region, σ1 is the standard deviation of the fault region, and σ2 is the standard deviation of the non-fault region.

[0032] Using the probability density function of the earthquake data, the first probability that each data point belongs to a fault is calculated;

[0033] The maximum likelihood attribute value of each data point is determined based on the parameter value of the data point corresponding to the first probability of maximum, and the maximum likelihood attribute of the strike-slip fault is determined by combining the maximum likelihood attribute value of each data point.

[0034] An automatic strike-slip fracture tracking system based on dynamic programming, comprising the steps of the aforementioned automatic strike-slip fracture tracking method based on dynamic programming, including:

[0035] The signal-to-noise ratio (SNR) attribute acquisition module is used to acquire the SNR attribute of strike-slip faults based on seismic data.

[0036] The data processing module is used to calculate the probability density function of the seismic data with the signal-to-noise ratio attribute of the strike-slip fault as a constraint, and to obtain the maximum likelihood attribute of the strike-slip fault based on the calculation result of the probability density function.

[0037] The fracture identification attribute acquisition module is used to use a dynamic programming algorithm to select the optimal path for the maximum likelihood attribute, and obtain the fracture identification attribute of the strike-slip fault based on the selected optimal path.

[0038] The interpretation and characterization module is used to perform geological interpretation on the fault identification attributes, characterize the occurrence of the strike-slip fault, and obtain the fault location, fault strike, and fault dip angle of the strike-slip fault.

[0039] An apparatus comprising:

[0040] Memory, used to store computer programs;

[0041] A processor is used to implement the steps of the above-described automatic tracking method for strike-slip fracture based on dynamic programming when executing the computer program.

[0042] A computer-readable storage medium storing a computer program, which, when executed by a processor, is used to implement the steps of the above-described automatic tracking method for strike-slip fracture based on dynamic programming.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] This invention also provides an automatic strike-slip fault tracking method based on dynamic programming. This method first estimates the signal-to-noise ratio (SNR) of seismic data to obtain the SNR attribute of the strike-slip fault. Then, using the SNR attribute as a constraint, it calculates the probability density function of the seismic data and obtains the maximum likelihood attribute of the strike-slip fault based on the calculation result. A dynamic programming algorithm is used to select the optimal path from the maximum likelihood attribute, thereby obtaining the fault identification attribute of the strike-slip fault. Finally, the fault identification attribute is geologically interpreted to characterize the attitude of the strike-slip fault, obtaining the fault location, strike, and dip angle. Compared with traditional methods, this method has stronger deterministic fault attribute extraction, provides more detailed and accurate characterization of the fault in cross-sections and planes, and effectively filters out the disappearance or misalignment of phase axes caused by non-faults. This method can clearly characterize the vertical distribution of strike-slip faults, thereby clearly determining the longitudinally developed segments of the strike-slip fault, with high reliability.

[0045] Preferably, in this invention, before obtaining the signal-to-noise ratio attribute of the strike-slip fault based on seismic data, the original seismic data is subjected to fault enhancement preprocessing. This method obtains high-precision seismic data by performing fault enhancement preprocessing on the original seismic data, which removes noise while highlighting fault characteristics; it reduces the serious interference of noise on the imaging accuracy of the fault and ensures the accuracy of the data.

[0046] More preferably, in this invention, the preset denoising method includes FK filtering, wavelet transform filtering, principal component analysis, and two-dimensional low-pass filtering, which improves the reliability of the denoising process.

[0047] More preferably, in this invention, the structural tensor is decomposed to obtain structural tensor eigenvalues; a diffusion tensor is calculated based on the structural tensor eigenvalues, and the original seismic data is diffuse-filtered using the diffusion tensor as a constraint to obtain seismic data. Here, the algorithm is improved based on traditional diffusion filtering. First, the original method's dip angle estimation result is greatly affected by noise, and dip angle correction is required when estimating the diffusion tensor, resulting in a large computational load. The improved method uses the structural tensor eigenvalues ​​to construct the diffusion tensor, reducing computational complexity and workload. Second, the improved method's diffusion tensor calculation combines the energy indication role of the first eigenvalue and the lateral difference indication role of the second eigenvalue of the structural tensor eigenvalue decomposition, thus enhancing the filtering result for strong-energy faults.

[0048] More preferably, in this invention, the three eigenvalues ​​are sorted in descending order, and the three eigenvalues ​​are represented as the first eigenvalue, the second eigenvalue, and the third eigenvalue according to the sorting result; the second eigenvalue is used as the eigenvalue of the structural tensor, and the eigenvector corresponding to the second eigenvalue is calculated; since seismic signals have the characteristics of large gradient changes in the time direction and small gradient changes in the spatial direction, the eigenvector corresponding to the second eigenvalue usually represents the direction of spatial gradient change; thus ensuring the accuracy of this method. Attached Figure Description

[0049] Figure 1 A flowchart of an automatic strike-slip fracture tracking method based on dynamic programming provided in an embodiment of the present invention;

[0050] Figure 2 A comparative schematic diagram of the improved diffusion filtering algorithm in an automatic strike-slip fracture tracking method based on dynamic programming provided in an embodiment of the present invention;

[0051] Figure 3 A comparison chart showing the intelligent identification effect of linear strike-slip faults provided in the embodiments of the present invention;

[0052] Figure 4 A flowchart of an automatic strike-slip fracture tracking method based on dynamic programming provided by the present invention;

[0053] Figure 5 This is a schematic diagram of the structure of an automatic strike-slip fracture tracking system based on dynamic programming provided by the present invention. Detailed Implementation

[0054] This invention provides an automatic strike-slip fracture tracking method based on dynamic programming, such as... Figure 4 As shown, it includes the following steps:

[0055] S0: The original seismic data is preprocessed with fault enhancement to obtain the seismic data.

[0056] The specific steps for enhancing preprocessing include:

[0057] The original seismic data was subjected to two-dimensional filtering using preset denoising methods to remove oblique interference noise. These preset denoising methods included FK filtering, wavelet transform filtering, principal component analysis, and two-dimensional low-pass filtering.

[0058] In three-dimensional space, the covariance or variance of the original seismic data after two-dimensional filtering is processed to obtain the structure tensor corresponding to the original seismic data; specifically, this step includes:

[0059] The raw seismic data is represented as a three-dimensional dataset F(x,y,z), where x,y,z represent spatial coordinates.

[0060] The original seismic data is cut into windows according to the preset cutting window size, and the covariance or variance of the original seismic data in each window is calculated to obtain the covariance matrix or variance matrix.

[0061] The covariance matrix within each window is averaged or accumulated to obtain the overall structural tensor of the original seismic data.

[0062] The structure tensor is subjected to eigenvalue decomposition to obtain its eigenvalues; the specific steps are as follows:

[0063] According to the eigenvalue decomposition formula, the structural tensor is decomposed into three eigenvalues; the eigenvalues ​​represent the magnitude of the gradient changes in the original seismic data in the spatial and temporal directions.

[0064] Sort the three eigenvalues ​​in descending order, and then represent them as the first eigenvalue, the second eigenvalue, and the third eigenvalue based on the sorting result.

[0065] The second eigenvalue is used as the eigenvalue of the structure tensor, and the corresponding eigenvector is calculated.

[0066] The diffusion tensor is calculated based on the eigenvalues ​​of the structural tensor, and the original seismic data is then subjected to diffusion filtering using the diffusion tensor as a constraint to obtain the seismic data.

[0067] S1: Obtain the signal-to-noise ratio of strike-slip faults based on seismic data;

[0068] S2: Using the signal-to-noise ratio attribute of the strike-slip fault as a constraint, calculate the probability density function of the seismic data, and obtain the maximum likelihood attribute of the strike-slip fault based on the calculation result of the probability density function; specific steps include:

[0069] The data points in the seismic data are represented as x. i , where i represents the index of the data point;

[0070] The signal-to-noise ratio of the strike-slip fault is used as a constraint condition for the Gaussian model, and the probability density function of the seismic data is calculated using the Gaussian model. The probability density function is expressed as follows:

[0071]

[0072] Wherein, P(x i |Fault) represents data point x i The probability of belonging to a fault, P(x) i (Non-fault) represents data point x i The probability of belonging to a non-fault region, u1 is the mean of the fault region, u2 is the mean of the non-fault region, σ1 is the standard deviation of the fault region, and σ2 is the standard deviation of the non-fault region.

[0073] Using the probability density function of the earthquake data, the first probability that each data point belongs to a fault is calculated;

[0074] The maximum likelihood attribute value of each data point is determined based on the parameter value of the data point corresponding to the first probability of maximum, and the maximum likelihood attribute of the strike-slip fault is determined by combining the maximum likelihood attribute value of each data point.

[0075] S3: Use dynamic programming algorithm to select the optimal path for the maximum likelihood attribute, and obtain the fracture identification attribute of the strike-slip fault based on the selected optimal path.

[0076] S4: Perform geological interpretation on the fracture identification attributes to characterize the occurrence of the strike-slip fault and obtain the fault location, fault strike, and fault dip angle of the strike-slip fault.

[0077] like Figure 5As shown, the present invention also provides an automatic strike-slip fault tracking system based on dynamic programming, comprising: a signal-to-noise ratio (SNR) attribute acquisition module, used to acquire the SNR attribute of the strike-slip fault based on seismic data; a data processing module, used to calculate the probability density function of the seismic data with the SNR attribute of the strike-slip fault as a constraint, and obtain the maximum likelihood attribute of the strike-slip fault based on the calculation result of the probability density function; a fault identification attribute acquisition module, used to use a dynamic programming algorithm to screen the optimal path of the maximum likelihood attribute, and obtain the fault identification attribute of the strike-slip fault based on the screened optimal path; and an interpretation and characterization module, used to perform geological interpretation of the fault identification attribute, characterize the attitude of the strike-slip fault, and obtain the fault location, fault strike, and fault dip angle of the strike-slip fault.

[0078] The present invention also provides an apparatus comprising: a memory for storing a computer program; and a processor for executing the computer program to implement the steps of the dynamic programming-based automatic tracking method for strike-slip fracture.

[0079] When the processor executes the computer program, it implements the above-mentioned steps for automatic tracking of strike-slip faults based on dynamic programming. For example: obtaining the signal-to-noise ratio (SNR) attribute of the strike-slip fault based on seismic data; calculating the probability density function of the seismic data using the SNR attribute of the strike-slip fault as a constraint, and obtaining the maximum likelihood attribute of the strike-slip fault based on the calculation result of the probability density function; using a dynamic programming algorithm to select the optimal path for the maximum likelihood attribute, and obtaining the fault identification attribute of the strike-slip fault based on the selected optimal path; performing geological interpretation on the fault identification attribute to characterize the attitude of the strike-slip fault, and obtaining the fault location, fault strike, and fault dip angle of the strike-slip fault.

[0080] Alternatively, when the processor executes the computer program, it implements the functions of each module in the above system, such as: a signal-to-noise ratio (SNR) attribute acquisition module, used to acquire the SNR attribute of the strike-slip fault based on seismic data; a data processing module, used to calculate the probability density function of the seismic data with the SNR attribute of the strike-slip fault as a constraint, and obtain the maximum likelihood attribute of the strike-slip fault based on the calculation result of the probability density function; a fault identification attribute acquisition module, used to use a dynamic programming algorithm to select the optimal path for the maximum likelihood attribute, and obtain the fault identification attribute of the strike-slip fault based on the selected optimal path; and an interpretation and characterization module, used to perform geological interpretation of the fault identification attribute, characterize the attitude of the strike-slip fault, and obtain the fault location, fault strike, and fault dip angle of the strike-slip fault.

[0081] For example, the computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing preset functions, wherein the instruction segments describe the execution process of the computer program in the dynamic programming-based automatic strike-slip fracture tracking device. For example, the computer program can be divided into a signal-to-noise ratio (SNR) attribute acquisition module, a data processing module, a fault identification attribute acquisition module, and an interpretation and characterization module. The specific functions of each module are as follows: the SNR attribute acquisition module is used to acquire the SNR attribute of the strike-slip fault based on seismic data; the data processing module is used to calculate the probability density function of the seismic data using the SNR attribute of the strike-slip fault as a constraint, and obtain the maximum likelihood attribute of the strike-slip fault based on the calculation result of the probability density function; the fault identification attribute acquisition module is used to use a dynamic programming algorithm to select the optimal path for the maximum likelihood attribute, and obtain the fault identification attribute of the strike-slip fault based on the selected optimal path; the interpretation and characterization module is used to perform geological interpretation of the fault identification attribute, characterize the attitude of the strike-slip fault, and obtain the fault location, fault strike, and fault dip angle of the strike-slip fault.

[0082] The automatic strike-slip fracture tracking device based on dynamic programming can be a computing device such as a desktop computer, laptop, handheld computer, or cloud server. This device may include, but is not limited to, processors and memory. Those skilled in the art will understand that the above examples of automatic strike-slip fracture tracking devices based on dynamic programming are not intended to limit the scope of such devices. They may include more components than described above, or combine certain components, or use different components. For example, the automatic strike-slip fracture tracking device based on dynamic programming may also include input / output devices, network access devices, buses, etc.

[0083] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor, or any conventional processor. The processor is the control center of the dynamic programming-based automatic strike-slip fracture tracking system, connecting all parts of the dynamic programming-based automatic strike-slip fracture tracking device via various interfaces and lines.

[0084] The memory can be used to store the computer program and / or modules. The processor implements various functions of the dynamic programming-based automatic strike-slip fracture tracking device by running or executing the computer program and / or modules stored in the memory and calling the data stored in the memory.

[0085] The memory may primarily include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a given function (such as sound playback or image playback). The data storage area may store data created based on the use of the mobile phone (such as audio data and phonebook entries). Furthermore, the memory may include high-speed random access memory and non-volatile memory, such as hard disks, RAM, plug-in hard disks, smart media cards (SMC), secure digital cards (SD), flash cards, at least one disk storage device, flash memory device, or other volatile solid-state storage devices.

[0086] The present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the described automatic tracking method for strike-slip fracture based on dynamic programming.

[0087] If the modules / units integrated in the dynamic programming-based automatic strike-slip fracture tracking system are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium.

[0088] Based on this understanding, the present invention can implement all or part of the processes in the above-mentioned automatic strike-slip fracture tracking method based on dynamic programming, or it can be accomplished by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium. When executed by a processor, the computer program can implement the steps of the above-mentioned automatic strike-slip fracture tracking method based on dynamic programming. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or a preset intermediate form, etc.

[0089] The computer-readable storage medium may include any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0090] It should be noted that the content contained in the computer-readable storage medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable storage medium does not include electrical carrier signals and telecommunication signals.

[0091] The present invention will be further described below with reference to embodiments and accompanying drawings:

[0092] Example 1

[0093] As described in the background section, carbonate oil and gas reservoirs are significantly controlled by strike-slip faults. Therefore, the identification and interpretation of strike-slip faults is both a key focus and a challenge in carbonate oil and gas reservoir management. In recent years, with the development of computer technology and the efforts of researchers both domestically and internationally, fault identification technology has made significant progress. Various fault identification methods have emerged, and fault interpretation has gradually evolved from simple manual identification to a combination of methods, greatly improving the accuracy of fault interpretation. Current fault identification technologies can be broadly categorized into two main types based on methodology:

[0094] (i) Fault identification technology based on seismic attributes: After processing the seismic data, relevant attributes are extracted, including some attributes that are more sensitive to faults (such as coherence, curvature, variance, maximum likelihood, etc.). ① Coherence volume technology is one of the better fault identification techniques currently available. It reflects the similarity of seismic data in the form of coherence attributes, highlighting the overall spatial development characteristics of faults, microfractures, lithology, or geological structures through coherence variations. Coherence volume technology can quantitatively identify faults from 3D seismic data, avoiding the influence of interpretation errors, thus greatly improving the accuracy of fault interpretation. Currently, coherence volume technology has evolved from the initial single-channel coherence to multi-channel coherence, then to eigenvalue coherence, and even to the coherence volume technology based on structure tensor and wavelet transform that emerged in the new century. The advantages of coherence volume fault identification are that it simplifies the process, requires fewer computational parameters, and provides good planar continuity. The disadvantage is that it lacks the ability to finely characterize the stage and grade of faults. ② Curvature analysis technology, with curvature attributes (maximum positive curvature, maximum negative curvature, Gaussian curvature, etc.), mainly predicts the location of fault development through the bending changes of seismic waveforms caused by faults. Its advantage is that it has high sensitivity to faults and can effectively identify faults of different scales, such as large, medium, and small. The disadvantages are that it is easily affected by the interpreted horizon and noise; ③ Variance volume analysis technology, which is a stratigraphic discontinuity detection technology based on probability variance analysis. When there is a fault underground, the reflection characteristics of the corresponding seismic trace in the seismic data volume will differ from those of adjacent seismic traces. The variance volume technology represents the difference in reflection characteristics of each seismic trace by calculating the variance between adjacent seismic traces, thereby completing the identification of the fault. The advantages are good vertical continuity and strong anti-interference ability. The disadvantage is that the identification effect of small faults is poor; ④ Maximum likelihood attribute is obtained by scanning the entire seismic data volume and calculating the similarity between data samples to obtain the most likely location and probability of fault development in the study area. The main steps include: seismic reflection characteristic analysis of faults, fault imaging enhancement under dip control, extraction of maximum likelihood attributes, and interpretation of attribute slices. Its advantages are strong ability to finely characterize faults and can characterize certain branch faults and internal structures of fault zones. The disadvantage is that it is affected by multiple interpretations, noise, and special geological bodies, resulting in a large number of false faults.

[0095] (II) Fault Interpretation Techniques Based on Tracking Algorithms. Automatic fault tracking and interpretation methods mainly include ant tracking and BP neural network tracking. ① Ant tracking technology: Its principle is to generate ant tracking paths by pre-setting fault conditions to characterize the spatial distribution features of faults. The ant tracking algorithm establishes a novel fault interpretation technique that highlights fault plane features. The advantages of this method are high tracking efficiency and high fault interpretation accuracy. The disadvantage is that it is greatly affected by lithological boundaries, causing numerous fault artifacts and making fault identification and fault combination difficult. ② BP neural network tracking technology: BP neural network tracking is a tracking algorithm established from an information processing perspective by learning the process of the human brain's neural network. It utilizes the high flexibility of neural network structures and the characteristics of seismic faults to propose a learning tracking technique. Its advantages are wide applicability and high fault plane tracking accuracy. The disadvantages are low computational efficiency, lack of convergence, and unstable learning tracking.

[0096] Currently, there are numerous methods for fault identification, but their effectiveness varies greatly, and they generally cannot meet the requirements for clear and intelligent tracking of strike-slip faults. Each technique has its own drawbacks or limitations. Further targeted technological research and development is needed for characterizing strike-slip faults in the Tarim Basin. The main problems and challenges currently facing strike-slip fault identification are as follows:

[0097] (1) At present, the plane resolution of the fault sensitive attribute is high. Although the fault distribution on the plane is relatively clear, the fault surface is difficult to be directly identified in the seismic data volume due to the small longitudinal displacement of the strike-slip fault.

[0098] (2) The development of faults is inherited and the superposition of multiple faults makes the development of faults diverse, and the traditional human-computer interaction interpretation is highly ambiguous.

[0099] (3) High-density fault planes have low human-computer interaction interpretation efficiency. At present, intelligent fault identification is still in the exploratory stage. How to reduce the multiple interpretations and improve interpretation efficiency through data processing and algorithm optimization is the key problem to be solved in fault identification.

[0100] (4) The development intensity of fractures is currently difficult to represent in the data volume; such as coherent enhancement, ant volume, likelihood, etc., the identified fractures are often difficult to distinguish the fracture order, and are affected by many non-fault factors, resulting in a relatively messy fracture characterization.

[0101] To address the aforementioned issues, this invention provides an automatic strike-slip fault tracking method based on dynamic programming. This method helps solve the problem that current strike-slip fault identification technologies struggle to directly identify fault surfaces and differentiate fault levels within seismic data volumes. It enables three-dimensional hierarchical sculpting of strike-slip faults, automatically generating interpretive cross-sectional data, and achieving automatic strike-slip fault tracking.

[0102] like Figure 1 As shown, this embodiment provides an automatic strike-slip fracture tracking method based on dynamic programming, which specifically includes the following steps:

[0103] S1: Acquire raw seismic data and perform fault enhancement preprocessing on the raw seismic data to obtain high-precision seismic data;

[0104] S2: Perform signal-to-noise ratio estimation on the high-precision seismic data to obtain the signal-to-noise ratio attributes of the strike-slip fault;

[0105] S3: Using the signal-to-noise ratio attribute of the strike-slip fault as a constraint, calculate the probability density function of the high-precision seismic data, and obtain the maximum likelihood attribute of the strike-slip fault based on the calculation result of the probability density function.

[0106] S4: Use dynamic programming algorithm to select the optimal path for the maximum likelihood attribute, and obtain the fracture identification attribute of the strike-slip fault based on the selected optimal path of the maximum likelihood attribute.

[0107] S5: Perform geological interpretation on the fracture identification attributes to characterize the occurrence of the strike-slip fault and obtain the fault location, fault strike, and fault dip angle of the strike-slip fault.

[0108] In step S2, the signal-to-noise ratio estimation of high-precision seismic data can be achieved using existing signal-to-noise ratio analysis techniques, such as the energy method and the power spectrum method, with the specific choice depending on the actual situation. In step S5, the geological interpretation of fault identification attributes can be achieved using the attribute slice analysis step in the maximum likelihood attribute analysis method.

[0109] Furthermore, in one embodiment, the present invention performs fault enhancement preprocessing on the original seismic data to obtain high-precision seismic data, including:

[0110] First, a preset denoising method is used to perform two-dimensional filtering on the original seismic data to remove oblique interference noise. This preset denoising method includes FK filtering, wavelet transform filtering, principal component analysis, and two-dimensional low-pass filtering.

[0111] Then, the covariance or variance of the original seismic data after two-dimensional filtering is processed in three-dimensional space to obtain the structural tensor corresponding to the original seismic data.

[0112] Perform eigenvalue decomposition on the structure tensor to obtain the structure tensor eigenvalues;

[0113] The diffusion tensor is calculated based on the eigenvalues ​​of the structural tensor, and the original seismic data is then subjected to diffusion filtering using the diffusion tensor as a constraint to obtain high-precision seismic data.

[0114] In practice, the signal-to-noise ratio of the raw seismic data is relatively low, and the noise in the seismic data includes random noise, oblique coherent signals, etc. The presence of this noise seriously interferes with the imaging accuracy of the fault, requiring further filtering and denoising. How to remove this noise while highlighting the fault imaging is the key to interpretive processing. There are various methods for noise removal, but in general, filtering is the main approach, such as bandpass filtering, two-dimensional filtering, and dip-guided filtering. However, filtering can also affect the imaging of the fault plane.

[0115] Specifically, existing technologies typically employ edge-preserving filtering for noise removal because conventional filtering processes can blur breakpoint information while improving the signal-to-noise ratio of the data. The purpose of edge-preserving filtering is to accurately identify breakpoint locations and perform filtering at non-fault locations, thereby highlighting fault features while removing noise.

[0116] Currently, edge-preserving filtering denoising algorithms are the "diffusion filtering algorithm" and the "eigenvalue decomposition algorithm". The principle of the diffusion filtering algorithm is as follows: input the original seismic data, obtain the structure tensor of the original seismic data, then calculate the dip angle of the strata, and calculate the diffusion coefficient along the dip angle direction. The diffusion coefficient is then used for diffusion filtering.

[0117] Among them, the diffusion filtering algorithm: adopts an iterative nonlinear diffusion filtering algorithm to achieve smoothing filtering in non-fault regions and enhanced filtering in fault regions.

[0118] When calculating the structure tensor and formation dip angle, the structure tensor is represented by T. Among them, g x : Gradient (in space) in the x-direction; g t : Gradient (time) in the t direction; λ u λ is the largest eigenvalue of the structure tensor decomposition. v : The smallest eigenvalue of the structural tensor decomposition; u: The eigenvector corresponding to the largest eigenvalue; v: The eigenvector corresponding to the smallest eigenvalue.

[0119] Vector u is perpendicular to the seismic reflection phase axis, while vector v is parallel to the seismic reflection phase axis. The dip angle of the formation is calculated using vector v.

[0120] The diffusion tensor and the structure tensor have the same eigenvector, which represents the direction of earthquake change. The dip angle of the strata can be indirectly obtained by using the angle between this direction of change and the original earthquake coordinates.

[0121] The principle behind eigenvalue decomposition (EVD) for denoising is as follows: elements in one space are linearly transformed to another space of the same dimension. Matrix eigenvalues ​​are measures of rotation and scaling of the eigenvectors in the original space, and the eigenvectors represent the new space. The significance of EVD lies in identifying the aspects in which a matrix produces the greatest dispersion, reducing overlap, and preserving more information.

[0122] However, both diffusion filtering and eigenvalue decomposition algorithms still have some drawbacks in denoising. Diffusion filtering relies on formation dip angle estimation and is sensitive to lateral differencing, which can easily amplify non-fault responses. The denoising effect of eigenvalue decomposition is affected by the size of the rectangular window.

[0123] like Figure 2 As shown, this invention addresses the shortcomings of the above-mentioned algorithms by improving the diffusion filtering algorithm. The main improvements are as follows:

[0124] 1. The original method's tilt angle estimation results are greatly affected by noise, and tilt angle correction is required when estimating the diffusion tensor, resulting in a large amount of computation. The improved method uses the eigenvalues ​​of the structure tensor to construct the diffusion tensor, which reduces the computational complexity and amount of computation.

[0125] 2. The improved diffusion tensor calculation method combines the energy indication role of the first eigenvalue and the lateral difference indication role of the second eigenvalue of the structural tensor eigendecomposition, thereby enhancing the filtering results for strong energy faults.

[0126] A larger diffusion coefficient indicates a stronger smoothness. The diffusion coefficient is obtained by normalizing and multiplying the two eigenvalues ​​of the structure tensor and refining them, resulting in a significant enhancement effect on faults at high energies. During the edge-preserving and denoising stage, it is still impossible to determine which edge is a fault and which is not, so indiscriminate protection is required at this stage. Filtering can improve the signal-to-noise ratio and remove some phase misalignments caused by noise.

[0127] Because seismic signals exhibit large temporal gradient changes and small spatial gradient changes during the extraction of fault information from 3D seismic data, the direction of the second eigenvalue of the structural tensor often corresponds to the direction of spatial gradient change. By utilizing the similarity between the second eigenvalue of the structural tensor eigenvalue decomposition and the fault trend, fault attributes can be extracted. The specific steps are as follows: First, extract structural tensor attributes from seismic data; second, perform eigenvalue decomposition on the structural tensor to obtain the second eigenvalue; third, perform spatial direction filtering on the seismic data to obtain fault attributes.

[0128] Based on the above improvements, this invention performs covariance or variance processing on the original seismic data after two-dimensional filtering in three-dimensional space to obtain the structure tensor corresponding to the original seismic data, including:

[0129] The raw seismic data is represented as a three-dimensional dataset F(x,y,z), where x,y,z represent spatial coordinates.

[0130] The original seismic data is cut into windows according to the preset cutting window size, and the covariance or variance of the original seismic data in each window is calculated to obtain the covariance matrix or variance matrix.

[0131] The covariance matrix within each window is averaged or accumulated to obtain the overall structural tensor of the original seismic data.

[0132] In practical application, the structural tensor is a second-order symmetric matrix, which can be calculated using methods such as variance and covariance matrices. For seismic data, the structural tensor in three-dimensional space can be represented as:

[0133]

[0134] Among them, E xx E xy E xz E yy E yz E zz These represent the variance or covariance of the seismic data in the three directions, respectively.

[0135] Assume the seismic data is a three-dimensional dataset, represented as F(x,y,z), where x, y, and z represent spatial coordinates. First, the seismic data is divided into windows according to a preset window size, which can be set to 3×3×3 or 5×5×5. Then, the covariance matrix or variance matrix is ​​calculated for the seismic data within each window.

[0136] For a variance matrix, the variance of each element can be calculated directly, for example:

[0137] E xx =var(F(x,y,z)) represents the variance in the x-direction;

[0138] E xy =cov(F(x,y,z),F(x,y,z+1)) represents the covariance in the x and y directions;

[0139] E yy =var(F(x,y,z+1)) represents the variance in the y-direction.

[0140] For a covariance matrix, the covariance of each element can be calculated, for example:

[0141] E xx =cov(F(x,y,z),F(x,y,z))

[0142] E xy =cov(F(x,y,z),F(x,y,z+1))

[0143] E yy =cov(F(x,y,z+1),F(x,y,z+1))

[0144] The calculation results within each window are summed or averaged to obtain the structural tensor of the entire seismic dataset.

[0145] By following the steps above, structural tensor properties can be extracted from seismic data to obtain a second-order symmetric matrix describing the gradient changes in the seismic data. This structural tensor can then be used for subsequent eigenvalue decomposition and fault information extraction.

[0146] Further, the structure tensor is subjected to eigenvalue decomposition to obtain the structure tensor eigenvalues, including:

[0147] According to the eigenvalue decomposition formula, the structural tensor is decomposed into three eigenvalues; the eigenvalues ​​represent the magnitude of the gradient changes in the original seismic data in the spatial and temporal directions.

[0148] Sort the three eigenvalues ​​in descending order, and then represent them as the first eigenvalue, the second eigenvalue, and the third eigenvalue based on the sorting result.

[0149] The second eigenvalue is used as the eigenvalue of the structure tensor, and the corresponding eigenvector is calculated.

[0150] In practice, eigenvalue decomposition of the structural tensor yields eigenvalues ​​and corresponding eigenvectors. Eigenvalues ​​represent the magnitude of the gradient change in a specific direction, while eigenvectors represent the direction of that gradient change. The eigenvalues ​​are sorted, typically with the first eigenvalue being the largest and the third the smallest. The eigenvector corresponding to the second eigenvalue is then determined. Since seismic signals exhibit large gradient changes in the time direction and small gradient changes in the spatial direction, the eigenvector corresponding to the second eigenvalue usually represents the direction of the spatial gradient change.

[0151] The mathematical calculation process for eigenvalue decomposition of the structure tensor T is as follows:

[0152] Construct a 3×3 identity matrix A.

[0153] A=T-λI

[0154] Where T is the structure tensor, I is the 3x3 identity matrix, and λ is the eigenvalue to be solved.

[0155] Calculate the determinant of matrix A:

[0156] det(A) = |T - λI|

[0157] Find the roots of det(A) = 0, i.e., the eigenvalues ​​λ1, λ2, λ3.

[0158] For each eigenvalue λ, substitute it into the equation A*v=0, where v is the eigenvector.

[0159] Solve the system of equations A*v=0 to obtain eigenvectors v1, v2, v3. Extract the eigenvector corresponding to the second eigenvalue, i.e., v2.

[0160] Furthermore, the direction of spatial gradient change can be determined based on the eigenvector v2 corresponding to the second eigenvalue. This eigenvector represents the direction of minimum spatial gradient change in seismic data, which is typically similar to the fault trend.

[0161] Spatial directional filtering methods, such as convolution operations, can be used to filter seismic data and highlight fault attributes. A suitable Sobel operator can be selected as the filter kernel. The specific steps are as follows:

[0162] Define a filter kernel or gradient operator, such as:

[0163]

[0164] Where g x and g y These represent the gradient operators in the x and y directions, respectively.

[0165] By convolving the filter kernel or gradient operator with the seismic data, the gradient components G in the x and y directions are obtained. x and G y .

[0166] Specifically, G x =g x *F(x,y,z), G y =g y *F(x,y,z). Where * represents the convolution operation, and F(x,y,z) represents the original seismic data.

[0167] Gradient components are filtered using specific filtering methods, such as maximum value filtering or average value filtering, to obtain fault attributes. The location and strike of the fault can be determined based on the filtered gradient magnitude or direction. Fault attributes can be extracted based on the filtering results.

[0168] Further analysis and extraction can be performed based on the filtered fault attributes, such as determining the fault location, fault strike, and fault dip angle.

[0169] Further, in one embodiment, the present invention uses the signal-to-noise ratio attribute of the strike-slip fault as a constraint to calculate the probability density function of the high-precision seismic data, and obtains the maximum likelihood attribute of the strike-slip fault based on the calculation result of the probability density function, including:

[0170] The data points in the high-precision seismic data are represented as x. i Where i represents the index of the data point; the signal-to-noise ratio attribute of the strike-slip fault is used as the constraint condition of the Gaussian model, and the probability density function of the high-precision seismic data is calculated using the Gaussian model. The probability density function is expressed as:

[0171]

[0172] Wherein, P(x i |Fault) represents data point x i The probability of belonging to a fault, P(x) i (Non-fault) represents data point x i The probability of belonging to a non-fault, u1 and u2 are the means of the fault and non-fault parts respectively, and σ1 and σ2 are the standard deviations of the fault and non-fault parts respectively;

[0173] Using the probability density function of the high-precision seismic data, the first probability that each data point belongs to a fault is calculated;

[0174] The maximum likelihood attribute value of each data point is determined based on the parameter value of the data point corresponding to the first probability of maximum, and the maximum likelihood attribute of the strike-slip fault is determined by combining the maximum likelihood attribute value of each data point.

[0175] Some progress has been made in fault identification technology both domestically and internationally, such as multichannel coherence technology, ant tracking technology, and maximum likelihood estimation. However, current conventional fault identification attributes are insufficient to accurately characterize strike-slip faults. Therefore, it is necessary to experiment with and develop fault identification attributes based on fault enhancement processing.

[0176] This invention employs a signal-to-noise ratio constrained fault extraction method. Based on the fault enhancement preprocessing described in the above embodiments, data analysis is performed to extract the fracture identification attributes of strike-slip faults, namely, the fault identification attribute body.

[0177] Because faults have a certain length of extension in the earthquake time direction, the signal-to-noise ratio (SNR) of the seismic response around the fault is relatively high, while the SNR of the seismic response caused by lateral lithological variations is relatively low. Maximum likelihood attributes have the advantage of enhancing fault continuity. However, since the original maximum likelihood attributes may also identify faults in some low SNR segments, using SNR attributes as constraints to extract maximum likelihood attributes and obtain fault attributes can effectively avoid this influence.

[0178] The process of extracting fault identification attributes using the signal-to-noise ratio constrained fault extraction method is as follows: input the original seismic data, perform denoising filtering using bandpass filtering or other methods, extract the signal-to-noise ratio attribute, set the signal-to-noise ratio threshold constraint to extract the maximum likelihood attribute, then determine the optimal path for the maximum likelihood attribute, and finally obtain the fault identification attribute body based on the maximum likelihood attribute of the optimal path.

[0179] In the specific implementation process, for fracture identification attributes, this invention employs an improved maximum likelihood attribute extraction method. Specifically, based on fault enhancement preprocessing, maximum likelihood attributes are extracted using signal-to-noise ratio constraints, followed by optimal path determination to identify the optimal path and obtain the fracture identification attributes. Compared to traditional methods, this method offers stronger determinism, provides more detailed and accurate characterization of fractures in cross-sections and planes, and effectively filters out non-fault-related phenomena such as the disappearance or misalignment of in-phase axes. The fracture identification attribute extraction process is as follows:

[0180] Step 1: Maximum Likelihood Attribute Extraction: Maximum likelihood attributes are extracted using the signal-to-noise ratio (SNR) constraint. The maximum likelihood attribute can be obtained by calculating the probability density function (PDF) of the data, typically fitted using a Gaussian model. Using the PDF, the probability that each data point belongs to a fault can be calculated.

[0181] In maximum likelihood attribute extraction, this can be achieved by calculating the probability density function (PDF) of the data. A common method is to use a Gaussian model for fitting, and the specific mathematical formula is as follows:

[0182] Assume the fault-enhanced preprocessed seismic data is x. i Where i represents the index of the data point, x i This is the value of the data point. Assume the fault portion of the earthquake data follows a Gaussian distribution, meaning the data points in the fault portion follow a Gaussian distribution with mean u1 and variance σ1; while the data points in the non-fault portion follow a Gaussian distribution with mean u2 and variance σ2.

[0183] The probability density function of earthquake data can then be expressed as:

[0184]

[0185] Wherein, P(x i |Fault) represents data point x i The probability of belonging to a fault, P(x) i (Non-fault) represents data point x i The probability of belonging to a non-fault region, u1 and u2 are the means of the fault and non-fault regions respectively, and σ1 and σ2 are the standard deviations of the fault and non-fault regions respectively.

[0186] Step 2: Optimal path determination: Based on the probability calculated from the maximum likelihood attribute, the location of the fault can be determined by determining the optimal path.

[0187] In dynamic programming algorithms for determining the optimal path, the following mathematical formula can be used:

[0188] Assuming there are N time points in the earthquake data and M candidate fault paths, define a matrix dp, where dp[i][j] represents the maximum likelihood attribute value when selecting the j-th path at the i-th time point. The dynamic programming process is as follows:

[0189] Initialization: dp[0][0] = log(P(0,0)), where P(0,0) represents the maximum likelihood attribute value at time 0 on path 0.

[0190] For the other dp[0][j], initialize to negative infinity, indicating that the j-th path was not selected at time point 0.

[0191] For the other dp[i][0], initialize it to negative infinity, which means that the 0th path will not be selected at any time point.

[0192] State transition: For each time point i (from 1 to N-1) and each path j (from 1 to M-1), update the value of dp[i][j]: dp[i][j] = max(dp[i-1][k] + log(P(i,j))), where k ranges from 0 to M-1, and P(i,j) represents the maximum likelihood attribute value at time point i on path j.

[0193] Optimal path selection: In the last row dp[N-1][j], find the column j* with the maximum value, i.e. max(dp[N-1][j]), then j* is the index of the optimal path.

[0194] After extracting the fracture identification attributes, the fracture identification attributes are interpreted in accordance with the processing method of the maximum likelihood attribute extraction method. That is, the fracture identification attributes are interpreted by attribute slices, which can be used to create a three-dimensional hierarchical engraving of the strike-slip fracture and automatically generate interpretive cross-sectional data, obtaining data such as the fault location, strike-slip fracture direction, and fracture mode of the strike-slip fault.

[0195] Example 2

[0196] In this embodiment, a typical strike-slip fault in the Lunnan block of the Tarim Basin is used as an example. The strike-slip fault is identified using the method in Embodiment 1 above, which further illustrates the technical advantages of the present invention.

[0197] The study area mainly develops three strike-slip fault systems: Late Caledonian, Late Hercynian, and Yanshanian. During the Late Caledonian, regional compression in the south was dominant, resulting in predominantly compressional-shear strike-slip faults. A compressional-shear fault zone formed by positive flower-like structures developed, along with a series of small short-axis anticlines. During the Late Hercynian, NE-trending extensional-shear faults developed along the early NE-trending fault zone, merging downwards with the main fault sections to reverse the fault characteristics. During the Yanshanian, along the main NE-trending strike-slip zone in the Mesozoic, small extensional-shear rift basins and en echelon structures developed, with strong extensional activity and predominantly right-lateral extensional strike-slip faults.

[0198] Based on the differences in the fault location, cutting relationship, and regional tectonic stress field background of the strike-slip faults, the strike-slip faults in the Golexi Block can be divided into three different structural styles from south to north: linear strike-slip faults, compressional-torsional braided strike-slip faults, and extensional braided strike-slip faults.

[0199] The Lunnan block was selected as the pilot area, with an area of ​​200 km². 2 A smart identification experiment was conducted on strike-slip faults. Based on the seismic data, a fault enhancement processing procedure was developed to highlight the reflection characteristics of faults in the seismic data volume. Then, an improved maximum likelihood estimation method was used to extract the fault identification attribute volume, and attribute thresholds were set for smart identification of strike-slip faults.

[0200] The fault characteristics of the test area were analyzed using the method described in the above embodiments as follows:

[0201] like Figure 3 As shown, Figure 3 This indicates the effectiveness of intelligent identification of linear strike-slip faults, where, Figure 3 (a) Partial attached diagrams show the original seismic profile; Figure 3 (b) Some of the attached figures show structurally guided filtering seismic profiles; Figure 3 (c) Some of the figures show the fault properties predicted by the method of the present invention; Figure 3 (d) Partial attached diagram shows the ant body profile. The linear model seismic response characteristics include: clear fault plane, obvious axial displacement in the same direction, but the difference in energy on the same reflection axis on both sides of the fault orientation when it is normal or reverse is typical of strike-slip faults.

[0202] Linear strike-slip faults appear as nearly vertical fault surfaces on seismic profiles. The fault planes are simple and clear, with obvious shifting of the same phase axis. The orientation sometimes resembles a normal fault and sometimes a reverse fault. There are energy differences on both sides of the fault surface, clearly demonstrating the characteristics of a strike-slip fault. From... Figure 3From the perspective of the intelligent identification effect of linear strike-slip faults, (a) is the original seismic profile; it can be seen that due to the low signal-to-noise ratio of seismic data, the interpretation accuracy of strike-slip faults is low; (b) is the profile after structural guidance filtering is performed on the original seismic profile, it can be seen that the seismic signal-to-noise ratio is significantly improved and the fault interpretation accuracy is significantly improved; (c) is the strike-slip fault attribute extracted by the method of the present invention, which can clearly characterize the vertical distribution of strike-slip faults, especially clearly determine the longitudinal development segment of strike-slip faults, and has high reliability; (d) is the traditional ant body profile, compared with the strike-slip fault attribute extracted by the method of the present invention, it can be seen that the traditional ant body has low accuracy in identifying strike-slip faults.

[0203] In summary, this invention provides an automatic strike-slip fault tracking method based on dynamic programming, which has the following advantages compared with existing strike-slip fault identification methods:

[0204] This invention provides an automatic strike-slip fault tracking method based on dynamic programming. The method obtains high-precision seismic data by performing fault enhancement preprocessing on raw seismic data, removing noise while highlighting fault characteristics. Based on this fault enhancement preprocessing, the method extracts maximum likelihood attributes using signal-to-noise ratio (SNR) constraints, then determines the optimal path to obtain fault identification attributes. Compared to traditional methods, this invention's fault attribute extraction method is more deterministic, providing a more detailed and accurate depiction of faults in cross-sections and planes, while effectively filtering out non-fault-related phenomena such as the disappearance or misalignment of phase axes. Furthermore, by using SNR constraints to extract maximum likelihood attributes, this invention avoids the influence of low SNR segments on strike-slip faults.

[0205] The above embodiments are merely one of the implementation methods for achieving the technical solution of the present invention. The scope of protection claimed by the present invention is not limited to this embodiment, but also includes any variations, substitutions and other implementation methods that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention.

Claims

1. An automatic strike-slip fracture tracking method based on dynamic programming, characterized in that, include: Obtain the signal-to-noise ratio of strike-slip faults based on seismic data; Using the signal-to-noise ratio attribute of the strike-slip fault as a constraint, the probability density function of the seismic data is calculated, and the maximum likelihood attribute of the strike-slip fault is obtained based on the calculation result of the probability density function. The optimal path is selected for the maximum likelihood attribute using a dynamic programming algorithm, and the fracture identification attribute of the strike-slip fault is obtained based on the selected optimal path. The geological interpretation of the fracture identification attributes is used to characterize the occurrence of the strike-slip fault and obtain the fault location, strike, and dip angle of the strike-slip fault.

2. The automatic tracking method for strike-slip fracture based on dynamic programming according to claim 1, characterized in that, Before obtaining the signal-to-noise ratio attribute of the strike-slip fault based on the seismic data, the original seismic data is subjected to fault enhancement preprocessing to obtain the seismic data.

3. The automatic tracking method for strike-slip fracture based on dynamic programming according to claim 2, characterized in that, The specific steps for fault enhancement preprocessing of the raw seismic data include: The original seismic data were subjected to two-dimensional filtering using a preset denoising method to remove oblique interference noise. In three-dimensional space, the covariance or variance of the original seismic data after two-dimensional filtering is processed to obtain the structural tensor corresponding to the original seismic data. Perform eigenvalue decomposition on the structure tensor to obtain the structure tensor eigenvalues; The diffusion tensor is calculated based on the eigenvalues ​​of the structural tensor, and the original seismic data is then subjected to diffusion filtering using the diffusion tensor as a constraint to obtain the seismic data.

4. The automatic tracking method for strike-slip fracture based on dynamic programming according to claim 3, characterized in that, The specific steps for performing covariance or variance processing on the original seismic data after two-dimensional filtering in three-dimensional space to obtain the structure tensor corresponding to the original seismic data include: The raw seismic data is represented as a three-dimensional dataset F(x,y,z), where x,y,z represent spatial coordinates. The original seismic data is cut into windows according to the preset cutting window size, and the covariance or variance of the original seismic data in each window is calculated to obtain the covariance matrix or variance matrix. The covariance matrix within each window is averaged or accumulated to obtain the overall structural tensor of the original seismic data.

5. The automatic tracking method for strike-slip fracture based on dynamic programming according to claim 3, characterized in that, The specific steps for performing eigenvalue decomposition on the structure tensor to obtain the eigenvalues ​​of the structure tensor include: According to the eigenvalue decomposition formula, the structural tensor is decomposed into three eigenvalues; the eigenvalues ​​represent the magnitude of the gradient changes in the original seismic data in the spatial and temporal directions. Sort the three eigenvalues ​​in descending order, and then represent them as the first eigenvalue, the second eigenvalue, and the third eigenvalue based on the sorting result. The second eigenvalue is used as the eigenvalue of the structure tensor, and the corresponding eigenvector is calculated.

6. The automatic tracking method for strike-slip fracture based on dynamic programming according to claim 3, characterized in that, The preset denoising methods include FK filtering, wavelet transform filtering, principal component analysis, and two-dimensional low-pass filtering.

7. The automatic tracking method for strike-slip fracture based on dynamic programming according to claim 1, characterized in that, The specific steps for calculating the probability density function of the seismic data using the signal-to-noise ratio attribute of the strike-slip fault as a constraint, and obtaining the maximum likelihood attribute of the strike-slip fault based on the calculation result of the probability density function, include: The data points in the seismic data are represented as x. i , where i represents the index of the data point; The signal-to-noise ratio of the strike-slip fault is used as a constraint condition for the Gaussian model, and the probability density function of the seismic data is calculated using the Gaussian model. The probability density function is expressed as follows: Wherein, P(x i |Fault) represents data point x i The probability of belonging to a fault, P(x) i (Non-fault) represents data point x i The probability of belonging to a non-fault region, u1 is the mean of the fault region, u2 is the mean of the non-fault region, σ1 is the standard deviation of the fault region, and σ2 is the standard deviation of the non-fault region. Using the probability density function of the earthquake data, the first probability that each data point belongs to a fault is calculated; The maximum likelihood attribute value of each data point is determined based on the parameter value of the data point corresponding to the first probability of maximum, and the maximum likelihood attribute of the strike-slip fault is determined by combining the maximum likelihood attribute value of each data point.

8. An automatic strike-slip fracture tracking system based on dynamic programming, used to implement the steps of the automatic strike-slip fracture tracking method based on dynamic programming as described in any one of claims 1-7, characterized in that, include: The signal-to-noise ratio (SNR) attribute acquisition module is used to acquire the SNR attribute of strike-slip faults based on seismic data. The data processing module is used to calculate the probability density function of the seismic data with the signal-to-noise ratio attribute of the strike-slip fault as a constraint, and to obtain the maximum likelihood attribute of the strike-slip fault based on the calculation result of the probability density function. The fracture identification attribute acquisition module is used to use a dynamic programming algorithm to select the optimal path for the maximum likelihood attribute, and obtain the fracture identification attribute of the strike-slip fault based on the selected optimal path. The interpretation and characterization module is used to perform geological interpretation on the fault identification attributes, characterize the occurrence of the strike-slip fault, and obtain the fault location, fault strike, and fault dip angle of the strike-slip fault.

9. A device, characterized in that, include: Memory, used to store computer programs; A processor, configured to implement the steps of the automatic tracking method for strike-slip fracture based on dynamic programming as described in any one of claims 1-7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it is used to implement the steps of the automatic tracking method for strike-slip fracture based on dynamic programming as described in any one of claims 1-7.

Images