A method for calculating fracture parameters based on prestack gather data

Abstract

This invention discloses a method for calculating fracture parameters based on pre-stack gather data, comprising the following steps: dividing the omnidirectional gather data into central angle gather data volumes, calculating the gradient and intercept attribute data volumes of each central angle gather data volume, and then correcting the gradient and intercept attribute data volumes; calculating the fracture intensity data volume and fracture direction data volume within different waveform ranges in each layer of the target layer in the study area; and performing data fusion processing on the fracture intensity data volume and fracture direction data volume within different waveform ranges in each layer of the study area in three-dimensional space to obtain the fracture intensity data volume and fracture direction data volume used for fracture prediction and evaluation in the study area. Compared with the prior art, the positive effect of this invention is that the fracture intensity and direction data volumes predicted by other fracture prediction methods are more accurate.

Description

Technical Field

[0001] This invention belongs to the field of petroleum geophysical exploration, specifically relating to a method for calculating fracture parameters based on pre-stack gather data. Background Technology

[0002] Fractures are crucial channels for the accumulation and migration of underground oil and gas. Fracture prediction refers to forecasting the development intensity of fractures or parameters related to fracture analysis. In most cases, fractures are tectonic fractures—fractures attributed to or associated with local tectonic events, including fault-related fracture systems, fracture systems related to stratigraphic uplift and arching, and fracture systems related to folds, etc. Southern marine carbonate reservoirs almost universally contain fractures. These fractures, under the dissolution action of acidic solutions, can further expand and extend, also serving to connect reservoirs, thus laying the foundation for the formation of high-quality carbonate reservoirs. Therefore, finding fractured carbonate reservoirs is one of the important objectives of marine oil and gas exploration.

[0003] Currently, there are essentially two methods for predicting underground fissures using seismic data—both related to pre-stack and post-stack seismic data. Additionally, geological empirical analysis techniques such as finite element analysis and tectonic stress field analysis are also used for fissure prediction. One method utilizes azimuth-based seismic stacking data, guided by the theory of azimuth anisotropy. To determine an ellipse with a center coordinate of (0, 0) and an arbitrary angle θ with the X-axis, three parameters need to be determined—the major axis radius *a*, the minor axis radius *b*, and θ. Therefore, determining an ellipse requires the coordinates of at least three known points. By calculating the ratios of the major and minor axes of these fitted ellipses, the fissure strength and direction can be calculated. Furthermore, the construction of fissure devices can be achieved through the combination of relevant alternative materials. Additionally, geophysical experimental techniques can be used to perform relevant calculations on the fissure devices and obtain their related seismic response parameters. Geophysical experimental techniques include seismic physical model testing techniques, rock physical testing and analysis techniques, etc. Among them, the seismic physical model experiment is a forward modeling method, which is to obtain the seismic physical model observation data under the condition of known geological seismic model; rock physical testing and analysis is to establish the relationship between rock physical properties and seismic response through rock core testing experiments, reveal the law of seismic wave propagation, and provide experience or theoretical basis for actual seismic response analysis and property inversion.

[0004] Numerous patents and technical documents exist regarding crack prediction, demonstrating that crack detection remains a hot research topic. Some patents and technical documents include: Fan Guozhang and Mou Yongguang's proposal to establish a crack distribution model and its seismic response (Fan Guozhang and Mou Yongguang's "Changes in P-wave Velocity in Anisotropic Crack Media and Their Influence on the Superposition of Common Centroid Gathers," published in *Petroleum Geophysical Exploration* in 2002); the invention titled "A Controllable Crack Parameter Physical Model and Its Fabrication Method" (Patent No.: 201210326132.4), which discloses a controllable crack parameter physical model and its fabrication method. This model is constructed by uniformly embedding crack-filling material into each background medium layer and stacking these layers; and the invention titled "A Seismic Physical Model, Its Preparation Method and Application" (Patent No.: 201010503831.2). Novel seismic physical model materials are synthesized using specialized equipment and processes. Seismic physical models are constructed using the principle of similarity to simulate actual geological structures, and are used for seismic wavefield and other research purposes. This allows for the study of the kinematic and dynamic characteristics of seismic wave propagation in complex areas, and also provides a relatively objective basis for verifying new methods and theories in oil and gas exploration and development. The patent titled "Fractured Prediction Method and Device" (Patent No.: 201010205983.4) proposes using a target layer time window to obtain the reflection amplitude of each seismic trace. Ellipse fitting is then performed using the obtained azimuth and reflection amplitude to predict the direction and density of fractures. However, these methods are quite difficult to implement and are easily affected by various factors, mainly presenting the following problems:

[0005] (1) Conventional fracture prediction technology often has a low match rate with well data when predicting small fractures or fractures in complex target layers, making prediction difficult.

[0006] (2) Using properties such as amplitude and frequency to calculate the direction and intensity of crack development by ellipse fitting, the prediction results often vary greatly, making it difficult to choose between these techniques and to judge their impact on the results. Summary of the Invention

[0007] To overcome the aforementioned shortcomings of existing technologies, this invention proposes a method for calculating two data volumes—crack intensity and direction—based on pre-stack gather data. This method employs a relatively simple technical process to address the problem of crack prediction, and more effectively calculates crack development parameters of the target layer in the relevant study area, thus effectively solving the problems existing in the aforementioned conventional crack prediction technologies.

[0008] The technical solution adopted by this invention to solve its technical problem is: a method for calculating crack parameters based on pre-stack gather data, comprising the following steps:

[0009] Step 1: Divide the all-round gather data into the central corner gather data volumes, and calculate the gradient and intercept attribute data volumes of each central corner gather data volume;

[0010] Step 2: After correcting the relevant gradient and intercept attribute data volumes, use them as input. Then, optimize the combination of two calculation models based on crack intensity and crack direction for each waveform category (based on layer segmentation and waveform classification) and the central angle attribute. Calculate two data volumes for parameters such as crack intensity and direction within the plane range of different waveform categories, and perform related three-dimensional spatial data volume fusion processing to obtain two data volumes for crack intensity and crack direction prediction in the target layer of the study area. The specific steps are as follows:

[0011] Step 201: Correct the gradient and intercept attribute data volumes of each central angle gather according to the relevant calculation formulas, and the resulting attribute data volumes will be used in the calculation of subsequent steps.

[0012] Step 202: Process the target layer within the study area according to the set relevant layer and waveform parameters to obtain different crack parameter calculation areas (establish a data area for calculating the same crack intensity and crack direction within the same classification and the same layer).

[0013] Step 203: For different crack parameter calculation areas, establish relevant crack strength and crack direction calculation models and input preferred attribute combinations, and calculate crack strength data volume and crack direction data volume within different waveform ranges in each layer of the target layer in the study area respectively.

[0014] Step 204: Perform data fusion processing on the crack strength data volume and crack direction data volume within different waveform ranges in each layer segment in three-dimensional space to obtain crack strength data volume and crack direction data volume for crack prediction and evaluation in the study area.

[0015] Compared with the prior art, the positive effects of the present invention are:

[0016] The method for calculating crack parameters based on pre-stack gather data provided by this invention can obtain accurate data volumes for both crack strength and crack direction compared to traditional crack calculation methods; and the crack strength and direction data volumes predicted by other crack prediction methods are more accurate. Attached Figure Description

[0017] The present invention will be described by way of example and with reference to the accompanying drawings, wherein:

[0018] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0019] A method for calculating crack parameters based on pre-stack gather data includes the following steps:

[0020] Step 1: Divide the omnidirectional gather data into data volumes for each central angle gather according to the set azimuth and incident angle range, and calculate the gradient and intercept attribute data volumes for each central angle gather data volume. The specific steps are as follows:

[0021] (1-1) The omnidirectional gather data volume is divided into multiple central angle gather data volumes according to the designed azimuth range, and the amplitude variation of each central angle is basically the same. Generally, the designed azimuth range is fixed, and the average value of the corresponding small azimuth data values ​​of each azimuth range is set as the relevant central angle data value. In actual operation, based on the set azimuth range, the number of central angle gathers, and the principle of ray symmetry, the omnidirectional gather data is divided into multiple set central angle gather data volumes within the azimuth range of 0°-180°. The omnidirectional gather data refers to the gather data after conventional field static correction, pre-stack denoising, amplitude compensation and deconvolution, residual static correction processing, and dynamic correction. Furthermore, the related stacking and migration processing are conventional seismic processing techniques and will not be described further in this invention. In principle, the more central angles designed, the higher the accuracy of crack strength calculation; conversely, the accuracy of crack strength calculation is relatively reduced. Secondly, the design of the central angles can be such that the increments between the central angles are equidistant or unequal, generally equidistant depending on the number of central angles between 0° and 180°. Generally, based on the symmetry principle of seismic rays, both the central angles and azimuth angles are set to gather data between 0° and 180°. Generally, the number of central angles divided in this invention must be greater than three. Specifically, ensuring that the amplitude variation of each central angle is basically consistent means that after performing conventional gather data processing such as amplitude preservation, fidelity preservation, and target layer time difference correction, the amplitude within each central angle gather is more easily calculated for AVO attributes.

[0022] (1-2) The common reflection point gather data corresponding to each central angle gather data are converted to incident angle gather data, and the converted central angle gather data are divided or removed according to the incident angle range with respect to the target layer, thus obtaining central angle-incident angle groups. The incident angle with respect to the target layer is generally not greater than 30°. Typically, the incident angle range of the set angle gather data is 8°-30°, and gather data with incident angles greater than 30° are removed (or discarded). In practice, the incident angle group can be set according to the incident angle range of the gather for the target layer in the depth domain, with the maximum incident angle being less than 30°. When calculating the incident angle range of the gather, the target layer depth data and layer velocity data volume of the relevant central angle CDP points can be used to perform angle conversion calculations on the relevant central angle gather data volumes according to relevant formulas, thus obtaining a series of central angle-incident angle groups. The target layer depth data can be obtained by calculating the layer position data and layer velocity data volume in the time domain, which will not be elaborated here. The layer velocity data volume and root mean square velocity data volume used for conversion can be calculated using relevant mature commercial software.

[0023] (1-3) Gradient and intercept attribute data are calculated for the common reflection point gather data corresponding to each of the divided central angle-incident angle groups, thereby obtaining the gradient and intercept attribute data volumes for each central angle-incident angle. The relevant attribute data volumes for gradient and intercept calculation are obtained by calculating the relevant gradient and intercept data based on the relationship between the seismic reflection amplitude and the incident angle (AVO) of the sampling points of the target layer of each central angle-incident angle gather data, thus obtaining two data volumes for each central angle: gradient and intercept. Generally, the calculation of gradient and attribute data volumes can be performed using relevant commercial software, and will not be elaborated upon here.

[0024] Preferably, the target interval refers to the interval containing reservoirs within the study area, which is the location of the interval for oil and gas exploration. Generally, the target interval refers to the strata between the geological strata containing oil and gas reservoirs within the study area. Specifically, based on the calibration results of the well-seismic composite records in relevant wells, the top and bottom stratigraphic data of the target interval are determined. After using the top and bottom stratigraphic data to perform stratigraphic locking, the two-way reflection time of the target interval is determined. In addition, based on the seismic reflection characteristics of the target interval, a seismic data interpretation grid is set, and stratigraphic data is interpreted throughout the entire study area to obtain relevant stratigraphic data.

[0025] Preferably, in the design of the relevant azimuth and central angle, the technical processing of this invention sets the direction of the observation system with true north as 0°, rotating clockwise for 360°. Based on the principle of symmetry, the 360° azimuth of the seismic data acquired in the field from the shot point to the receiver point is converted into 180° azimuth angles. For a given azimuth angle range, its central angle is calculated, and the central angle represents the defined azimuth angle range. The calculation formula is as follows:

[0026]

[0027] In the formula, θ i For the central angle of the i-th gather data in the design, To determine the minimum azimuth angle for the i-th gather data in the design, Let i be the maximum azimuth angle of the i-th gather data in the design, where i ≥ 2.

[0028] Preferably, the dynamically corrected CMP gather data is converted to the angular domain to obtain angular gather data. Furthermore, the omnidirectional CMP gather data can be further processed to improve signal-to-noise ratio, resolution, and fidelity. Specifically, the angular gather conversion can be performed using one of the following calculation formulas, depending on the actual situation. The relevant angular gather conversion formulas can be one of the following two:

[0029]

[0030] In the formula, θ is the incident angle of the angular path in the first case, x is the gun-receiver distance, v is the root mean square velocity, and t0 is the two-way travel time at zero offset.

[0031]

[0032] In the formula, α is the incident angle of the angular path in the second case, and v int For layer velocity, v rms Let t be the root mean square velocity, t be the two-way travel time, and x be the gun-receiver distance.

[0033] Step 2: After correcting the relevant gradient and intercept attribute data volumes, use them as input. Then, optimize the combination of two calculation models based on crack intensity and crack direction for each waveform category (based on layer segmentation and waveform classification) and central angle attribute. Calculate two data volumes for parameters such as crack intensity and direction within the plane range of different waveform categories, and perform related three-dimensional space data volume fusion processing to obtain two data volumes for crack intensity and crack direction prediction in the target layer of the study area. The specific steps are as follows:

[0034] (2-1) The gradient and intercept data volumes of each central angle are corrected, and the calculation models and central angle combinations for fracture intensity and direction of each classified waveform after layer segment and waveform classification are optimized to determine the relevant fracture calculation model and the optimized central angle combination attribute data volume. The fracture calculation model described in this invention refers to establishing various functional relationships for calculating fracture intensity and direction using measured data from wells. These functional relationships can be BP neural network regression mathematical models and their related improvements. Furthermore, sample wells and blind wells are established, and the gradient and intercept attributes (after correction) calculated by the relevant fracture calculation model and a series of optimized central angle attribute combinations are used as input attributes for the relevant calculation model through elliptic fitting. Additionally, waveform classification in this step refers to identifying different seismic waveforms based on the reflection waveform characteristics of the target layer segment using supervised or unsupervised classification methods and classifying them into different categories designed after classification. Generally, classification methods can include automatic waveform classification based on SOM neural networks, automatic waveform classification based on hierarchical clustering, automatic waveform classification based on probabilistic model clustering, or supervised waveform classification based on deterministic probability distribution (EM). These techniques are already implemented in a large number of commercial software programs.

[0035] Preferably, the process of performing segmentation and waveform classification on the target layer to obtain different waveform classification planes within different segments is described, and the data range of each waveform classification plane is optimized using two calculation models for crack intensity and crack direction, as well as a combination of central angles. Specifically, the steps for performing segmentation and waveform classification on the target layer to obtain different waveform classification planes within different segments, and for optimizing the data range of each waveform classification plane using two calculation models for crack intensity and crack direction, are as follows:

[0036] a) Perform segmentation calculations on the target segment according to the set fixed time window parameters and number of segments to obtain a series of segmented segments. Generally, the time window of the segmented segments should be greater than or equal to three-quarters of the wavelength. In practice, this can be determined based on the segmented segment test results, expert experience, and crack prediction accuracy. In principle, the more segments divided, the higher the crack calculation accuracy; conversely, the crack calculation accuracy is relatively lower. In principle, the segmented time thicknesses are equal, but different time thicknesses can be divided according to relevant actual conditions. If the target segment (e.g., segment thickness not greater than three-quarters of the wavelength) does not meet the requirements for segmentation, this step can be ignored, and the next step regarding waveform classification can be directly performed.

[0037] (b) Using layered seismic data, after setting the relevant number of waveform classifications for different layers, waveform classification processing is performed within each layer to obtain the waveform classification plane data range for different layers. In practice, for a specific layer, the known well logging data and post-stack impedance inversion methods and parameters can be used to perform impedance inversion on the 3D post-stack seismic data volume to obtain the relevant post-stack impedance data volume; then, the relevant waveform classification methods and parameters are used to classify the waveforms of the impedance inversion data volume for that layer, thus obtaining the relevant waveform classification plane data. This process is repeated to complete different waveform classifications for different layers. In principle, the number of waveform classifications for each layer can be the same or different. The determination of the number of classifications should be based on the actual situation of the seismic data, expert experience, and fracture prediction accuracy. Generally, the designed number of waveform classifications should be such that there are at least two drilled wells within each waveform plane range after the relevant waveform classification. This is relatively beneficial for fracture parameter modeling for each waveform classification.

[0038] c) Optimize the calculation models for fracture intensity and fracture direction within the data range of each waveform classification plane. In practice, this mainly involves establishing a series of calculation models for fracture intensity and direction within the relevant waveform classification plane, based on known measured fracture intensity and direction data from wells and calculated fracture intensity and direction data from each central angle (after correction). These models are then optimized to obtain the preferred calculation models for fracture intensity and direction within each waveform classification data. In this invention, fracture direction refers to the dominant fracture direction at fracture development points in the well—the primary fracture direction, which has a single directional property and is also called the first fracture direction. The 3D post-stack seismic data volume is obtained by stacking and migrating the omnidirectional angle gather data volume.

[0039] d) Based on the preferred fracture parameter calculation model, optimize the relevant central angle combinations. In practice, set the relevant combination number increment (generally set to one), start the calculation with a combination number of three, and use the preferred fracture parameter calculation model and the measured fracture strength and direction data in the blind well to optimize the calculation results of different combinations. The combination with the smallest average error or the combination with the second smallest error is selected as the preferred central angle attribute combination, or the combination with the smallest average error is selected as the preferred central angle attribute combination.

[0040] Preferably, the gradient and intercept properties of each central angle are calculated, and the gradient and intercept properties of each central angle are corrected. The corrected gradient and intercept properties of each central angle are used to establish the input attribute test and optimal central angle combination properties of the relevant crack calculation model.

[0041] Preferably, the gradient properties of the relevant sampling points at each central corner are corrected to obtain the corrected gradient properties for the next step. The formula for calculating the corrected gradient properties is as follows:

[0042]

[0043] In formula (1), The corrected gradient attribute of the i-th sampling point at the j-th central angle. The intercept attribute data value of the i-th sampling point of the j-th central angle. The gradient attribute data value of the i-th sampling point at the j-th central angle. It is the average of the sum of the intercept data values ​​of each central angle at the sampling point.

[0044] Preferably, the intercept attributes of the relevant sampling points at each central angle are corrected to obtain the corrected intercept attributes for the next step. The formula for calculating the corrected intercept attributes is as follows:

[0045]

[0046] In formula (2), The corrected intercept attribute of the i-th sampling point of the j-th central angle. The wave impedance attribute data value of the i-th sampling point at the j-th central angle. The gradient attribute data value of the i-th sampling point at the j-th central angle. This is the average of the sum of the wave impedance attribute data values ​​of each central angle at the sampling point. Generally, the main steps for obtaining the wave impedance attribute data values ​​of each central angle gather are to superimpose and migrate the data volumes of each central angle gather using the same method and parameters to obtain the post-stack data volume of each central angle; then, using well logging data and stratigraphic data to model and perform seismic inversion on the post-stack data volumes of each central angle, the relevant wave impedance data volumes of each central angle are obtained.

[0047] Preferably, in the above-mentioned gradient and intercept correction processing, the absolute values ​​of the gradient and intercept should be calculated before the correction processing is performed.

[0048] Preferably, given that the crack parameters predicted in this invention are mainly crack strength and crack direction, the establishment of the relevant crack parameter calculation model and the main operation steps are as follows:

[0049] (a) Based on the relationship between each central angle and gradient or intercept attribute (after correction), establish vector patterns for different central angle attributes. That is, establishing in the two-dimensional polar coordinate plane in, Let be the seismic attribute (either gradient or intercept attribute) data value for the i-th central angle. This represents the central angle corresponding to the seismic attribute. Polar coordinates are established primarily for elliptical fitting calculations to determine crack strength and direction. The polar coordinate system uses true north as 0° and rotates clockwise by 360°.

[0050] (b) Establishment of the relevant crack strength calculation model. This crack strength calculation model mainly consists of relevant mathematical functional relationships. The mathematical functional relationships for crack strength calculation are as follows:

[0051] F s i =f(K i m (3)

[0052] In formula (3), F s i Let f be the crack strength data value of the i-th sampling point on the s-th CDP point, f be the functional relationship for calculating crack strength, and K be the crack strength data value of the s-th sampling point. i m To optimize the central angle combination, a certain attribute is used to calculate the m-th crack strength data value based on ellipse fitting. Where K... i m The acquisition of data values ​​is mainly based on the gradient or intercept properties of the optimal combination of central angles. After performing ellipse fitting calculations on the relevant data projection points in polar coordinates, the crack strength data value K for this combined attribute is obtained. i m and crack direction θ i m These two data values ​​are then used in the calculation of crack strength and direction. Ellipse fitting for crack strength and direction calculation is a well-established technique and will not be elaborated upon here. In practice, based on the number of central angles, a preferred combination of central angle attributes for ellipse fitting calculation is set to determine the gradient or intercept attributes of each central angle within this combination. Perform relevant ellipse fitting calculations to obtain the corresponding crack strength data value K. i m and crack direction θ i m Data. Similarly, the crack strength data value K is calculated by optimizing the combination of various central angle settings for gradient and intercept attributes. i m and crack direction θ i mThis yields a series of crack strength and crack direction data values. Generally, the preferred number of central angle combinations is set to 4-6.

[0053] The formula for calculating the crack direction is as follows:

[0054] θ s i =f(θ) i m (4)

[0055] In formula (4), θ s i Let f be the crack direction data value of the i-th sampling point on the s-th CDP point, f be the functional relationship for calculating the crack direction, and θ be the crack direction data value. i m To optimize the central angle combination, the data value of the m-th crack direction is calculated based on ellipse fitting.

[0056] In practice, for example, within a specific waveform classification range of a certain layer segment, the measured fracture intensity of relevant sampling points is established based on the fracture-free and fractured segments within that layer segment in each well. and data values ​​of crack direction Alternatively, relevant curve data, and based on the gradient and intercept seismic attributes of each central angle corresponding to its relevant sampling points, after testing and optimization of relevant central angle attribute combinations, an elliptic fitting calculation is performed on the optimized d (d≥3) central angle attribute combinations to obtain a series of fracture strength and direction data as input and training. This establishes two functional relationships for calculating fracture strength and direction, which together form the fracture parameter calculation model. The correspondence between the measured fracture strength and direction in the well and the calculated fracture strength data is mainly based on the time-depth relationship in the well. The measured fracture strength curve in the well is transformed from the depth domain into the time domain and resampled for calculation, thus obtaining the correspondence between the fracture strength in the well and the calculated fracture strength. Similarly, the correspondence between the measured fracture direction and the calculated fracture direction data can also be established.

[0057] Preferably, the related combinations, such as for gradient attributes, refer to taking d combinations of gradient attribute data from n central angle gradient attributes each time, regardless of their order, and combining them into a group. The formula for calculating the number of combinations is as follows:

[0058]

[0059] In formula (5), n is the number of central angles, and d≥3 is set, n>d. In actual operation, the size of the number of d combinations for testing can be determined according to the crack prediction accuracy, test conditions, expert experience, etc., or a series of d1, d2, d3, etc. combinations can be set. Generally, the number of d combinations can be set to an incremental mode (e.g., an increment of 1) to test the crack calculation model and optimize the central angle combination attributes.

[0060] (c) The calculation models for fracture intensity and fracture direction within each waveform classification range after relevant layering and waveform processing, and the optimization of combinations of central angle attribute data, are then performed on the optimized calculation models and the ellipse fitting calculation results using the central angle combinations for subsequent calculations. In practice, the optimization of the calculation model involves establishing sample wells and blind wells for known wells. Sample wells are primarily used to establish calculation models for fracture intensity or direction, while blind wells are used to test and optimize these models. In practice, the relevant test models and input attributes (fracture intensity and direction data calculated by ellipse fitting after setting the central angle combination) can be used to calculate the fracture intensity and direction data for blind wells. Error analysis or correlation coefficient calculation is performed between this data and the measured fracture intensity and direction values ​​of the target layer in the blind wells. The fracture intensity and fracture direction calculation models corresponding to the smallest average error or largest correlation coefficient for each blind well are selected for the next step. In addition, for the optimal selection of central angle combinations, the main operation steps are to set a series of combinations with incremental characteristics in the number of combinations, to perform elliptical fitting calculations on the fracture strength and direction of the blind well using the set series of combinations, to perform error analysis or correlation coefficient calculations on the obtained results and the measured fracture strength and fracture direction data of the target layer in the blind well, and to select the central angle combination corresponding to the minimum average error or the maximum correlation coefficient of each blind well to proceed to the next step.

[0061] Preferably, the steps for optimizing the calculation model for crack strength and crack direction, and for optimizing the combination of central angle attribute data, can be performed in any order. In practice, a crack calculation model can be set first, and the relevant central angle attribute combinations can be tested and optimized. Then, the optimized central angle attribute combinations can be used to test and optimize a series of crack calculation models. Alternatively, relevant central angle attribute combinations can be set first, the crack calculation model can be tested and optimized, and then the optimized crack calculation model can be used to test and optimize a series of central angle attribute combinations to determine the relevant optimized central angle attribute combinations.

[0062] (2-2) Using the crack strength and direction calculation models and their corresponding preferred central angle combinations within each waveform after different waveform classifications of each layer segment as input, two data volumes for crack parameter calculation within the waveform range are obtained through relevant calculations. This process is repeated to complete the calculation of crack strength and direction data volumes within each layer segment and its different waveform classification ranges, resulting in relevant crack strength and direction data volumes. Specifically, the crack strength and crack direction data are obtained by optimizing and combining the corrected gradient and intercept attributes of each central angle, followed by ellipse fitting calculations to obtain their respective crack strength and crack direction data volumes, thus obtaining a series of two types of crack strength and crack direction data volumes. These are then used as input data, and after calculation by the preferred crack strength calculation model, a crack strength data volume (input crack strength data volume) is obtained; after calculation by the preferred crack direction calculation model, a crack direction data volume (input crack direction data volume) is obtained.

[0063] (2-3) The crack intensity and crack direction data volumes of different waveform classifications within each layer segment are fused in three-dimensional space to obtain the crack intensity data volume and crack direction data volume for the entire study area regarding crack prediction and evaluation. The three-dimensional fusion processing of different data volumes can be implemented using mature commercial software and will not be described in detail in this invention.

[0064] The prediction results of the method of the present invention were verified by the following embodiments:

[0065] like Figure 1 As shown, this example demonstrates the calculation of fracture intensity and direction data volumes for a marine target layer in a three-dimensional work area, based on the technical process of this invention. In oil and gas exploration in this area, it was found that fracture prediction of the reservoir is crucial; wells drilled into areas with well-developed fractures within the reservoir have been tested and found to yield high-yield industrial gas flows.

[0066] The central angle, azimuth range, and incident angle range related to the setting of the omnidirectional gather in step one are determined. In practice, based on the actual seismic data of the area and expert experience, the number of central angles is determined to be 18, namely 5°, 15°, 25°, 35°, 45°, 55°, 65°, 75°, 85°, 95°, 105°, 115°, 125°, 135°, 145°, 155°, 165°, and 175°. The central angles are designed with equal increments. The azimuth range of the relevant central angles is set to ±5° of the central angle, and the incident angle of the main target layer is set to be less than 30°. The set incident angle range is 5°-30°. Based on the set azimuth and incident angle range and other relevant parameters, the omnidirectional gather data is divided and cut according to the azimuth and incident angle range parameters to obtain gather data volumes of each central angle-incident angle common reflection point.

[0067] Secondly, using relevant well data, synthetic record calibration and geological data, and 3D post-stack seismic data, the target stratigraphic levels in the study area were determined. After manual interpretation of the stratigraphic data based on the interpretation grid, the interpreted stratigraphic data underwent interpolation and smoothing to obtain the stratigraphic data used for subsequent calculations. Furthermore, according to the needs of subsequent steps, omnidirectional angle gathers and azimuth and incident angle gathers were stacked and migrated to obtain 3D post-stack seismic data volumes and post-stack seismic data volumes for each central angle. These data volumes were then used for subsequent impedance inversion calculations.

[0068] In this step, the gradient and intercept of the target layer related sampling points are calculated for a total of 18 center angle-incident angle co-reflection point gather data volumes to obtain the related center angle-incident angle gradient and intercept attribute data volumes.

[0069] In step two, after correcting the gradient and intercept data volumes according to the relevant correction formulas, and designing the relevant number of layers and time windows, the target layer is layered to obtain each layer. Then, waveform classification is performed on each layer to obtain the waveform classification range for different layers. Based on the test results, expert experience, and crack prediction accuracy of the crack strength and direction calculation model for the relevant waveform classification range, a crack parameter calculation model for crack strength and direction is established and optimized. Then, the relevant central angle attribute combinations are optimized. Using the gradient and intercept attributes of the optimized combination number, ellipse fitting calculations are performed to obtain crack strength data volumes and crack direction data volumes. These are then input into the relevant crack strength and direction calculation models to calculate the crack strength and direction data volumes for each layer and each classification waveform. Finally, the calculated crack strength and direction data volumes are fused in three-dimensional space to obtain the crack strength and crack direction data volumes for crack prediction in the study area. In practice, for the target section of the study area, the corrected gradient and intercept attribute data, and the optimal number of central angle combinations (6 and 8) after testing based on the central angle attribute data, are used to calculate the fracture intensity and fracture direction with respect to the elliptic fitting of the gradient and intercept, respectively. These two types of data volumes are then used as inputs to the optimal fracture parameter calculation model. In practice, given that the time thickness of the target layer in the study area is less than three-quarters of the wavelength (the bottom of the target layer and a 40ms upward time window), it is not layered; instead, waveform classification is directly applied. The designed number of waveform classifications is 6, resulting in six different waveform ranges. The number of wells within these ranges meets the requirements for modeling the relevant fracture parameters. Furthermore, based on the test results of relevant blind wells within a specific waveform classification area, expert experience, and fracture prediction accuracy, a feedforward backpropagation (BP) neural network model was chosen as the calculation model for fracture intensity and direction. The BP neural network model was trained using measured fracture intensity and direction data from relevant sampling points in the target layer above the well, along with their corresponding central angle attributes for the optimized combination, calculated via elliptic fitting. This resulted in two sets of BP neural network models for subsequent fracture intensity and direction calculations. This process was repeated to optimize the calculation models for fracture intensity and direction within the six waveform classifications. In practice, for a given waveform classification, the fracture intensity and direction data volume calculated using the optimized central angle attribute combination was used as input. After calculation by the optimized fracture intensity and direction calculation model, two data volumes (fracture intensity data volume and fracture direction data volume) were obtained. This process was repeated to calculate the six fracture intensity data volumes and six fracture direction data volumes within the six waveform classifications. The two data volumes were then fused in three-dimensional space to obtain one fracture intensity data volume and one fracture direction data volume.In practice, for training the fracture strength calculation model, the fracture strength parameters of each well within a certain waveform classification range are used as the training data for the neural network. The network model structure is 2-9-2. After 17,800 iterations, the sum of squared systematic errors is 0.003, which is less than the expected error of 10. -2 Upon completion of the training, a preferred crack strength calculation model within the waveform classification range is obtained. Similarly, the training of calculation models for crack strength and crack direction within each waveform classification is completed, resulting in a preferred crack parameter calculation model.

[0070] The fracture parameter prediction results of this invention, when compared and analyzed with subsequent drilling data in the study area, showed a consistency rate of over 81.5%, meeting the geological requirements.

[0071] The above technical solution is only one embodiment of the present invention. For those skilled in the art, based on the application methods and principles disclosed in the present invention, it is easy to make various types of improvements or modifications, and not limited to the methods described in the above specific embodiments of the present invention. Therefore, the methods described above are only preferred and are not restrictive.

[0072] All parts not covered in this invention are the same as or can be implemented using existing technologies.

Claims

1. A method for calculating crack parameters based on pre-stack gather data, characterized in that: Includes the following steps: Step 1: Divide the all-round gather data into the central corner gather data volumes, and calculate the gradient and intercept attribute data volumes of each central corner gather data volume; Step 2: Correct the gradient and intercept attribute data volumes of each central corner gather; Step 3: Calculate the crack intensity data volume and crack direction data volume within different waveform ranges in each layer of the target layer in the study area; Step 4: Perform data fusion processing on the crack strength data volume and crack direction data volume within different waveform ranges in each layer segment in three-dimensional space to obtain crack strength data volume and crack direction data volume for crack prediction and evaluation in the study area. The method described in step one for dividing the omnidirectional gather data into central angle gather data volumes and calculating the gradient and intercept attribute data volumes of each central angle gather data volume is as follows: 1) Divide the omnidirectional gather data into multiple central angle gather data volumes according to the designed azimuth range, and make the amplitude variation of each central angle consistent; 2) Perform incident angle gathering transformation on the common reflection point gathering data corresponding to each central angle gathering data, and divide or cut the transformed central angle gathering data according to the incident angle range with respect to the target layer to obtain the central angle-incident angle group; 3) Calculate the gradient and intercept attribute data of the common reflection point gather data corresponding to each central angle-incident angle group, so as to obtain the gradient and intercept attribute data of each central angle gather data volume; The gradient attribute data volume of each central corner gather data volume is corrected according to the following formula: In the formula, The corrected gradient attribute of the i-th sampling point at the j-th central angle. The intercept attribute data value of the i-th sampling point of the j-th central angle. The gradient attribute data value of the i-th sampling point at the j-th central angle. It is the average of the sum of the intercept data values ​​of each central angle at this sampling point; The intercept attribute data volume of each central angle gather data volume is corrected according to the following formula: In the formula, The corrected intercept attribute of the i-th sampling point of the j-th central angle. The wave impedance attribute data value of the i-th sampling point at the j-th central angle. The gradient attribute data value of the i-th sampling point at the j-th central angle. It is the average of the sum of the wave impedance attribute data values ​​of each center angle at the sampling point.

2. The method for calculating crack parameters based on pre-stack gather data according to claim 1, characterized in that: The number of central angles is greater than three.

3. The method for calculating crack parameters based on pre-stack gather data according to claim 1, characterized in that: The incident angle range of the target layer is 8 o -30 o .

4. The method for calculating crack parameters based on pre-stack gather data according to claim 1, characterized in that: Step 3 describes the method for calculating the crack intensity data volume and crack direction data volume within different waveform ranges in each layer of the target layer in the study area: 1) The target layer of the study area is divided into layers to obtain each layer; 2) Perform waveform classification processing on each layer segment to obtain the waveform classification range of the relevant different layers segment; 3) Establish and optimize the calculation model for crack strength and crack direction; 4) Utilize the gradient and intercept attributes of the optimal combination number of the central angle attribute to perform ellipse fitting calculations to obtain the crack strength data volume and crack direction data volume; 5) Input the crack strength data volume and crack direction data volume into the preferred crack strength and crack direction calculation model respectively, and calculate the crack strength data volume and crack direction data volume within different waveform ranges in each layer segment.

5. The method for calculating crack parameters based on pre-stack gather data according to claim 4, characterized in that: The crack strength calculation model consists of the following mathematical functional relationships: F s i = f 1 (K i m ) In the formula, F s i The crack strength data value is the i-th sampling point on the s-th CDP point. f 1 K is the functional relationship for calculating crack strength. i m To optimize the central angle combination, the m-th crack strength data value is calculated based on ellipse fitting.

6. The method for calculating crack parameters based on pre-stack gather data according to claim 5, characterized in that: The crack direction calculation model consists of the following mathematical function relationship: s i = f 2 （ i m ） In the formula, s i For the crack direction data value of the i-th sampling point on the s-th CDP point, f 2 Here is the functional relationship for calculating the crack direction. i m To optimize the central angle combination, the data value of the m-th crack direction is calculated based on ellipse fitting.

7. The method for calculating crack parameters based on pre-stack gather data according to claim 4, characterized in that: The optimal number of combinations for the central angle attribute is 4-6.

Images