Multi-domain Small-Scale Fracture Identification Method and Related Equipment Based on Probabilistic Neural Networks

Abstract

This invention discloses a multi-domain small-scale fracture identification method based on probabilistic neural networks. By performing edge detection on the "three instantaneous" attributes of seismic records, it fully mines fracture information from attributes such as amplitude, frequency, and phase. Then, probabilistic neural network technology is used for multi-domain small-scale fracture optimization and detection, ultimately achieving accurate identification of small-scale fractures. This innovation improves the speed and quality of small-scale fracture identification, enhances the accuracy and reliability of micro-fracture prediction, provides important evidence for the integration of geological and seismic data, and lays a solid foundation for subsequent comprehensive seismic interpretation, dynamic and static characteristic analysis of oil and gas reservoirs, and adjustment of oil and gas reservoir development plans.

Description

Technical Field

[0001] This invention belongs to the field of earthquake comprehensive interpretation technology, specifically a multi-domain small-scale fracture identification method and related equipment based on probabilistic neural networks. Background Technology

[0002] Fractures are not only seepage systems in low-permeability reservoirs but also important sites for oil and gas enrichment and storage. Accurate identification of small-scale fracture systems is a crucial research area in the exploration and development of complex oil and gas reservoirs. Small-scale fracture systems are relatively small in size, making identification and prediction difficult, and represent a global research challenge. Conducting identification of micro-fracture systems and accurately predicting their spatial distribution patterns plays a vital role in improving exploration success rates and oil and gas reservoir development efficiency.

[0003] Many methods are commonly used in the industry for fault prediction, including 3D seismic attribute prediction technology based on post-stack attributes, multi-wave multi-component prediction technology based on shear wave splitting phenomena, and P-wave azimuth anisotropy prediction technology based on P-wave kinematics and dynamics. However, these methods suffer from problems such as high computational cost, low resolution and accuracy of small-scale fault prediction, ambiguity of fault strike and dip, and significant susceptibility to noise, failing to provide strong support for oil and gas exploration and development. In recent years, deep learning-based fault identification technology has developed rapidly, using convolutional neural networks, probabilistic neural networks, residual neural networks, and other methods to extract information such as fault planes and strikes at medium and large scales, achieving automatic fault identification and interpretation. Summary of the Invention

[0004] This invention provides a multi-domain small-scale fracture identification method and related equipment based on probabilistic neural networks, which solves the problems of existing methods such as large computational load, low resolution and accuracy of small-scale fracture prediction, ambiguity of fracture strike and dip, and great influence of noise, thus failing to provide strong support for oil and gas exploration and development.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A multi-domain small-scale fracture recognition method based on probabilistic neural networks has the following characteristics:

[0007] Transform the seismic record data to obtain the instantaneous amplitude, instantaneous phase, and instantaneous frequency data of the earthquake;

[0008] By performing edge detection on the instantaneous amplitude, instantaneous phase, and instantaneous frequency data of the earthquake, multi-domain fault information is obtained;

[0009] Multi-domain fracture information is input into a trained multi-domain small-scale fracture recognition model to obtain fracture data volume.

[0010] Preferably, the process of obtaining the multi-domain small-scale fracture identification model is as follows:

[0011] Randomly generate fracture models with different lengths, dip angles, fault displacements, and seismic noise levels that conform to the actual conditions of the work area;

[0012] Seismic records were obtained based on fault models;

[0013] Transforming seismic records yields instantaneous amplitude, instantaneous phase, and instantaneous frequency attributes, allowing for the acquisition of various information from seismic records in both the time and frequency domains.

[0014] The transverse variation information is obtained from the instantaneous amplitude, instantaneous phase, and instantaneous frequency attributes to characterize the fracture.

[0015] The instantaneous amplitude, instantaneous phase, and instantaneous frequency attributes of edge detection are used as training data, and the forward fracture model is used as the label. 70% of the data is set as the training set and 30% as the validation set. The neural network model is used for training, and the training is completed when the accuracy of the validation set reaches 90%.

[0016] Preferably, the neural network model is a probabilistic neural network.

[0017] Preferably, the instantaneous amplitude, instantaneous phase, and instantaneous frequency data of the earthquake are obtained by using Hilbert transform.

[0018] Preferably, the specific process of obtaining the instantaneous amplitude, instantaneous phase, and instantaneous frequency data of an earthquake using Hilbert transform is as follows: first, the earthquake record is transformed into an analytical signal, and then the instantaneous amplitude, instantaneous phase, and instantaneous frequency data of the earthquake are defined through the analytical signal;

[0019] The transformation formula is:

[0020] s(t)=s r (t)+js i (t),

[0021] Where s i (t),s r (t) represents the result after the Hilbert transform;

[0022] The formulas for defining the instantaneous amplitude, instantaneous phase, and instantaneous frequency data of an earthquake by analyzing the signal are as follows:

[0023]

[0024]

[0025]

[0026] Among them, A i (t), ψ i (t), f i(t) represents the instantaneous amplitude, instantaneous phase, and instantaneous frequency, respectively.

[0027] Preferably, the Canny edge detection algorithm is used to obtain information on the lateral change of the instantaneous amplitude, instantaneous phase, and instantaneous frequency attributes.

[0028] Preferably, the specific method for obtaining seismic records based on the fault model is to perform a convolution operation between the generated fault model and the seismic wavelet of the work area to obtain a synthetic seismic record.

[0029] A multi-domain small-scale fracture recognition system based on probabilistic neural networks includes:

[0030] Transformation module: Transforms seismic record data to obtain instantaneous amplitude, instantaneous phase, and instantaneous frequency data of the earthquake;

[0031] Detection module: It performs edge detection on the instantaneous amplitude, instantaneous phase, and instantaneous frequency data of the earthquake to obtain multi-domain fracture information;

[0032] Recognition module: Inputs multi-domain fracture information into a trained multi-domain small-scale fracture recognition model to obtain fracture data volume.

[0033] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the multi-domain small-scale fracture identification method based on a probabilistic neural network.

[0034] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the multi-domain small-scale fracture identification method based on a probabilistic neural network.

[0035] Compared with existing technologies, this invention has the following advantages: This invention provides a multi-domain small-scale fracture identification method based on probabilistic neural networks. By performing edge detection on the "three instantaneous" attributes of seismic records, it fully mines fracture information from attributes such as amplitude, frequency, and phase. Then, probabilistic neural network technology is used for multi-domain small-scale fracture optimization and detection, ultimately achieving accurate identification of small-scale fractures. This innovation improves the speed and quality of small-scale fracture identification, enhances the accuracy and reliability of micro-fracture prediction, provides important evidence for the integration of geological and seismic data, and lays a solid foundation for subsequent comprehensive seismic interpretation, dynamic and static characteristic analysis of oil and gas reservoirs, and adjustment of oil and gas reservoir development plans. Attached Figure Description

[0036] Figure 1 This is a flowchart of the multi-domain small-scale fracture recognition method based on probabilistic neural networks of the present invention;

[0037] Figure 2 This is a block diagram of the multi-domain small-scale fracture recognition system based on probabilistic neural networks of the present invention;

[0038] Figure 3 This is a flowchart of the multi-domain small-scale fracture recognition technology based on probabilistic neural networks in an embodiment of the present invention.

[0039] Figure 4 It is a comparison chart of synthetic earthquake records and "three instantaneous" attributes.

[0040] Figure 5 It is a fracture information map of edge detection with "three instants" attributes.

[0041] Figure 6 The results of multi-domain small-scale fault identification are shown in the image (light color represents the seismic discontinuity detection volume, characterizing the fault development zone), where a is the coherence volume and seismic overlay image, and b is the small-scale fault detection volume and seismic overlay image of this method. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0043] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0044] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0045] In the description of the embodiments of the present invention, it should be noted that if terms such as "upper," "lower," "horizontal," or "inner" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of the invention is in use, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. Furthermore, terms such as "first" and "second" are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0046] Furthermore, the use of the term "horizontal" does not imply that the component must be absolutely horizontal, but rather that it can be slightly tilted. For example, "horizontal" simply means that its direction is more horizontal than "vertical," and does not mean that the structure must be completely horizontal, but can be slightly tilted.

[0047] In the description of the embodiments of the present invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in the present invention according to the specific circumstances.

[0048] like Figure 1 As shown, this invention provides a multi-domain small-scale fracture recognition method based on a probabilistic neural network, comprising:

[0049] S101 transforms seismic record data to obtain instantaneous amplitude, instantaneous phase, and instantaneous frequency data of the earthquake;

[0050] S102 uses edge detection to obtain multi-domain fracture information by processing the instantaneous amplitude, instantaneous phase, and instantaneous frequency data of the earthquake.

[0051] S103 inputs multi-domain fracture information into a trained multi-domain small-scale fracture recognition model to obtain fracture data volume.

[0052] The process of obtaining the multi-domain small-scale fracture recognition model is as follows:

[0053] Randomly generate fracture models with different lengths, dip angles, fault displacements, and seismic noise levels that conform to the actual conditions of the work area;

[0054] Seismic records were obtained based on fault models;

[0055] Transforming seismic records yields instantaneous amplitude, instantaneous phase, and instantaneous frequency attributes, allowing for the acquisition of various information from seismic records in both the time and frequency domains.

[0056] The transverse variation information is obtained from the instantaneous amplitude, instantaneous phase, and instantaneous frequency attributes to characterize the fracture.

[0057] The instantaneous amplitude, instantaneous phase, and instantaneous frequency attributes of edge detection are used as training data, and the forward fracture model is used as the label. 70% of the data is set as the training set and 30% as the validation set. The neural network model is used for training, and the training is completed when the accuracy of the validation set reaches 90%.

[0058] The neural network model used is a probabilistic neural network.

[0059] The instantaneous amplitude, instantaneous phase, and instantaneous frequency data of the earthquake were obtained by using Hilbert transform.

[0060] The specific process of obtaining the instantaneous amplitude, instantaneous phase, and instantaneous frequency data of an earthquake using Hilbert transform is as follows: first, the earthquake record is transformed into an analytical signal, and then the instantaneous amplitude, instantaneous phase, and instantaneous frequency data of the earthquake are defined through the analytical signal;

[0061] The transformation formula is:

[0062] s(t)=s r (t)+js i (t),

[0063] Where s i (t),s r (t) represents the result after the Hilbert transform;

[0064] The formulas for defining the instantaneous amplitude, instantaneous phase, and instantaneous frequency data of an earthquake by analyzing the signal are as follows:

[0065]

[0066]

[0067]

[0068] Among them, A i (t), ψ i (t), f i (t) represents the instantaneous amplitude, instantaneous phase, and instantaneous frequency, respectively.

[0069] The Canny edge detection algorithm is used to extract information on the lateral changes from instantaneous amplitude, instantaneous phase, and instantaneous frequency attributes.

[0070] The specific method for obtaining seismic records based on the fault model is to perform a convolution operation between the generated fault model and the seismic wavelet of the work area to obtain a synthetic seismic record.

[0071] Another embodiment of the present invention provides a multi-domain small-scale fracture recognition method based on a probabilistic neural network, such as... Figure 3 As shown:

[0072] Step 1: Randomly generate fracture models with different lengths, dip angles, fault displacements, and seismic noise levels that conform to the actual conditions of the work area;

[0073] Step 2: Perform convolution operation between the generated fault model and the seismic wavelet of the work area to obtain the synthetic seismic record;

[0074] Step 3: Perform Hilbert transform on the seismic record to obtain the "three instantaneous" attributes (i.e., instantaneous amplitude, instantaneous phase, and instantaneous frequency), thereby acquiring various information about the seismic record in the time and frequency domains.

[0075] Let the input signal be s(t), and the result after the Hilbert transform be s i (t),s r (t), the corresponding analytic signal is:

[0076] s(t)=s r (t)+js i (t), (1)

[0077] The three instantaneous properties of the analytic signal can be defined as follows:

[0078]

[0079] Among them, A i (t), ψ i (t), f i (t) represents the instantaneous amplitude, instantaneous phase, and instantaneous frequency, respectively.

[0080] Step 4: Use the Canny edge detection algorithm on the "three instants" attribute to obtain the lateral change information to characterize the fracture;

[0081]

[0082] Where G(x,y) and Θ(x,y) represent the magnitude and direction of the gradient.

[0083] Step 5: Use the "three instantaneous" attributes of edge detection as training data and the forward fracture model as the label. Set 70% as the training set and 30% as the validation set. Use a probabilistic neural network for training. Training is complete when the accuracy of the validation set reaches 90%.

[0084] Probabilistic neural networks (PNNs) are a branch of radial basis function networks and are often used for pattern classification. They integrate density function estimation and Bayesian decision theory into traditional radial basis function networks, making the training process simple and the convergence rapid.

[0085] Probabilistic Neural Networks (PNNs) are neural network models based on statistical principles. Essentially, they are parallel algorithms developed based on the Bayesian minimum risk criterion, representing a completely forward computation process. They have short training times, are less prone to local optima, and offer high classification accuracy. Regardless of the complexity of the classification problem, with sufficient training data, an optimal solution under the Bayesian criterion can be guaranteed. Therefore, this paper uses probabilistic neural networks for intelligent optimization of three-dimensional spatial morphology to reduce the ambiguity in identifying minute fractures.

[0086]

[0087] Where p is the probability matrix of fracture, and R is the matrix related to the fracture direction.

[0088] Step Six: Process the actual earthquake record data according to Steps Three and Four, and use it as input to identify small fractures in the work area using a trained probabilistic neural network. Finely characterize their location, length, dip angle, fault displacement, and other distribution features to achieve the purpose of small-scale fracture identification.

[0089] 1. Figure 3 This is a flowchart of a multi-domain small-scale fracture identification technique based on probabilistic neural networks. This method primarily uses multi-domain attribute edge detection and probabilistic neural networks for small-scale fracture identification. First, seismic records are synthesized using a fracture model established through forward modeling, and their "three instantaneous" attributes are calculated to obtain various information from the seismic records in the time and frequency domains. Then, the Canny edge detection algorithm is used on the "three instantaneous" attributes to obtain fracture information. Finally, a probabilistic neural network is used to identify minute fractures within the work area, precisely characterizing the distribution features of small-scale fractures.

[0090] 2. Figure 4 It is a comparison chart of synthetic seismic records and "three instantaneous" attributes. Instantaneous amplitude and instantaneous phase are less affected by noise and can better reflect the characteristics of fractures in seismic records. Instantaneous frequency is more affected by noise and has a poorer effect on characterizing fractures.

[0091] 3. Figure 5 These are the results of fracture identification with different attributes. The "three instantaneous" attributes can generally reflect the characteristics of fractures in seismic records well, and contain multi-domain (time domain and frequency domain) information, expanding the amount of fracture information in seismic records and enabling better extraction of details from small-scale fractures. Compared with the fracture information from the "three instantaneous" attributes, the multi-domain small-scale fracture identification results map has higher fracture accuracy, clearer and more continuous fracture details, effectively suppressed noise, and a more realistic and natural spatial distribution, laying a good foundation for small-scale fracture prediction.

[0092] 4. Figure 6 This is a comparison of the results of small-scale fault seismic identification in the actual work area. Compared with the traditional post-stack coherent volume fault identification technology, the new technology has higher accuracy in identifying small-scale faults, with clearer and more continuous details. The new technology is significantly better than the traditional coherent volume technology in identifying micro-scale faults.

[0093] like Figure 2 As shown, the present invention also provides a multi-domain small-scale fracture recognition system based on a probabilistic neural network, characterized in that it includes:

[0094] Transformation module: Transforms seismic record data to obtain instantaneous amplitude, instantaneous phase, and instantaneous frequency data of the earthquake;

[0095] Detection module: It performs edge detection on the instantaneous amplitude, instantaneous phase, and instantaneous frequency data of the earthquake to obtain multi-domain fracture information;

[0096] Recognition module: Inputs multi-domain fracture information into a trained multi-domain small-scale fracture recognition model to obtain fracture data volume.

[0097] An embodiment of the present invention provides a terminal device. This terminal device includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps in the various method embodiments described above. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the various device embodiments described above.

[0098] 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.

[0099] The terminal device may be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0100] The processor may 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.

[0101] The memory can be used to store the computer program and / or module. The processor implements various functions of the terminal device by running or executing the computer program and / or module stored in the memory and calling the data stored in the memory.

[0102] If the modules / units integrated into the terminal device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0103] Although embodiments of the present invention have been described above in conjunction with the accompanying drawings, the present invention is not limited to the specific embodiments and application fields described above. The specific embodiments described above are merely illustrative and instructive, and not restrictive. Those skilled in the art, guided by the specification, can make many other modifications without departing from the scope of the claims of the present invention, and all of these modifications are within the scope of protection of the present invention.

Claims

1. A multi-domain small-scale fracture recognition method based on probabilistic neural networks, characterized by: Transform the seismic record data to obtain the instantaneous amplitude, instantaneous phase, and instantaneous frequency data of the earthquake; By performing edge detection on the instantaneous amplitude, instantaneous phase, and instantaneous frequency data of the earthquake, multi-domain fault information is obtained; Multi-domain fracture information is input into a trained multi-domain small-scale fracture recognition model to obtain fracture data volume; The process of obtaining the multi-domain small-scale fracture recognition model is as follows: Randomly generate fracture models with different lengths, dip angles, fault displacements, and seismic noise levels that conform to the actual conditions of the work area; Seismic records were obtained based on fracture models; Transforming seismic records yields instantaneous amplitude, instantaneous phase, and instantaneous frequency attributes, allowing for the acquisition of various information from seismic records in both the time and frequency domains. The transverse variation information is obtained from the instantaneous amplitude, instantaneous phase, and instantaneous frequency attributes to characterize the fracture. The instantaneous amplitude, instantaneous phase, and instantaneous frequency attributes of edge detection are used as training data, and the forward fracture model is used as the label. 70% of the data is set as the training set and 30% as the validation set. The neural network model is used for training. The training is completed when the accuracy of the validation set reaches 90%. The characteristic feature is that the neural network model adopts a probabilistic neural network.

2. The multi-domain small-scale fracture recognition method based on probabilistic neural networks according to claim 1, characterized in that, The instantaneous amplitude, instantaneous phase, and instantaneous frequency data of the earthquake were obtained by using Hilbert transform.

3. The multi-domain small-scale fracture recognition method based on probabilistic neural networks according to claim 2, characterized in that, The specific process of obtaining the instantaneous amplitude, instantaneous phase, and instantaneous frequency data of an earthquake using Hilbert transform is as follows: first, the earthquake record is transformed into an analytical signal, and then the instantaneous amplitude, instantaneous phase, and instantaneous frequency data of the earthquake are defined through the analytical signal; The transformation formula is: in This is the result after the Hilbert transform; The formulas for defining the instantaneous amplitude, instantaneous phase, and instantaneous frequency data of an earthquake by analyzing the signal are as follows: in, These represent instantaneous amplitude, instantaneous phase, and instantaneous frequency, respectively.

4. The multi-domain small-scale fracture recognition method based on probabilistic neural networks according to claim 1, characterized in that, The Canny edge detection algorithm is used to obtain information on the lateral changes of instantaneous amplitude, instantaneous phase, and instantaneous frequency attributes.

5. The multi-domain small-scale fracture recognition method based on probabilistic neural networks according to claim 1, characterized in that, The specific method for obtaining seismic records based on the fault model is to perform a convolution operation between the generated fault model and the seismic wavelet of the work area to obtain a synthetic seismic record.

6. A multi-domain small-scale fracture recognition system based on probabilistic neural networks, characterized in that, The method for implementing the multi-domain small-scale fracture recognition method based on probabilistic neural networks according to any one of claims 1-5 includes: Transformation module: Transforms seismic record data to obtain instantaneous amplitude, instantaneous phase, and instantaneous frequency data of the earthquake; Detection module: It performs edge detection on the instantaneous amplitude, instantaneous phase, and instantaneous frequency data of the earthquake to obtain multi-domain fracture information; Recognition module: Inputs multi-domain fracture information into a trained multi-domain small-scale fracture recognition model to obtain fracture data volume.

7. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the multi-domain small-scale fracture identification method based on probabilistic neural networks as described in any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the multi-domain small-scale fracture identification method based on probabilistic neural networks as described in any one of claims 1 to 5.

Images