Seismic facies diagram obtaining method and device

Abstract

The invention discloses a seismic facies diagram obtaining method. The method comprises the following steps: obtaining a target fusion attribute stratum slice; establishing an expert experience recognition label according to the target fusion attribute stratum slice; and obtaining a seismic facies diagram according to the expert experience cognition label and a deep convolutional neural network deep learning technology of a U-NET structure. The method is used for depicting the seismic facies of the target layer of the shallow water limited terrace environment in the research work area, the beach nucleus, the beach wing, the tidal channel and the lagoon facies are effectively identified, and the problems that the beach nucleus and the tidal channel are difficult to distinguish and part of the beach nucleus and the beach wing are overlapped in the prior art are solved. The research results provide technical conditions for later geological modeling (especially facies modeling) research.

Description

Technical Field

[0001] The present invention relates to the technical field of petroleum geophysical exploration, and in particular to a method and device for obtaining a seismic phase map. Background Art

[0002] Seismic facies quantitative classification technology is the foundation of reservoir phase-controlled inversion and geological modeling, especially phase modeling. Since the 1970s, seismic facies technology has evolved from qualitative analysis based on seismic reflection structures to quantitative identification based on pattern recognition technology.

[0003] However, because seismic facies identification technology relies on quantitative learning and classification based on the similarity of input seismic attribute values, two challenges remain in its current practical application: ① When attribute values ​​of some facies zones are similar, pattern recognition techniques have difficulty distinguishing these zones; and ② due to the influence of the surrounding rock mass, the seismic reflection characteristics or attribute values ​​of some facies zones may not share common characteristics, making them difficult to identify. This creates significant uncertainty for subsequent reservoir prediction and geological modeling. Summary of the Invention

[0004] The purpose of the present invention is to provide a method and device for obtaining a seismic phase map, which realizes the human-computer interactive quantitative identification of the seismic phase of the bioclastic beach reservoir, and provides more accurate technical support and lateral trends for development geological research.

[0005] To achieve the above object, the present invention provides a method for obtaining a seismic phase map, the method comprising:

[0006] Obtain target fusion attribute stratum slice;

[0007] Establish expert experience recognition labels based on the target fusion attribute stratum slice;

[0008] The seismic phase map is obtained by using the deep convolutional neural network deep learning technology based on the expert experience recognition labels and the U-NET structure.

[0009] Optionally, obtain target fusion attribute stratum slices, including:

[0010] Get 3D attribute body;

[0011] Extracting stratigraphic slices with different attributes from the 3D attribute volume based on the target layer top interpretation layer and the target layer bottom interpretation layer;

[0012] Determine multiple target seismic attributes by combining the target layer's planar sedimentary facies map and reservoir knowledge;

[0013] RGB color fusion is performed on the multiple target seismic attributes to obtain target fusion attribute stratum slices.

[0014] Optionally, obtain 3D attribute bodies, including:

[0015] Obtain 3D seismic data volume;

[0016] A 3D attribute volume is obtained according to the 3D seismic data volume.

[0017] Optionally, the 3D attribute body includes: amplitude, frequency, phase, and coherence.

[0018] Optionally, based on the target fusion attribute stratum slice, expert experience recognition labels are established, including:

[0019] According to the plane sedimentary facies map of the target layer and the plane characteristics of the attributes, different facies belts are interpreted on the target fused attribute stratigraphic slice to obtain expert experience recognition labels.

[0020] Optionally, obtaining a seismic phase map based on the expert experience recognition labels and a deep convolutional neural network deep learning technique using a U-NET structure includes:

[0021] Obtain the probability distribution of each phase based on the expert experience recognition labels and the deep convolutional neural network deep learning technology of the U-NET structure;

[0022] The facies belts are merged sequentially to obtain the seismic facies map.

[0023] Optionally, based on the expert experience recognition labels and a deep convolutional neural network deep learning technique with a U-NET structure, a probability distribution of each phase is obtained, including:

[0024] For the target facies belt, the deep convolutional neural network deep learning technology of the U-NET structure is used to extract features of the target fusion attribute stratum slice to obtain the smallest image unit that reflects the characteristics of the original input data;

[0025] Based on the expert experience recognition labels, the feature data labeled by the expert experience is set as the final learning goal and quality control standard, and the smallest image unit is used as the training object for iterative training to obtain the final data expression that highlights the characteristics of the target phase band;

[0026] According to the final data expression of the target phase belt characteristics, the probability distribution of each phase is obtained.

[0027] Optionally, according to the final data expression of the target phase belt characteristics, a probability distribution of each phase is obtained, including:

[0028] Using a self-organizing neural network technique to perform geometric distance weighting processing on the final data expression, to obtain the distribution probability of the target phase belt;

[0029] Repeat the test for other phase zones to obtain the probability distribution for each phase.

[0030] The present invention also provides a device for obtaining a seismic phase map, the device comprising:

[0031] An acquisition unit, used for obtaining a target fusion attribute stratum slice;

[0032] An establishing unit, configured to establish an expert experience recognition label according to the target fusion attribute stratum slice;

[0033] A learning unit is used to obtain a seismic phase map based on the expert experience recognition label and the deep convolutional neural network deep learning technology of the U-NET structure.

[0034] Optionally, obtain target fusion attribute stratum slices, including:

[0035] Get 3D attribute body;

[0036] Extracting stratigraphic slices with different attributes from the 3D attribute volume based on the target layer top interpretation layer and the target layer bottom interpretation layer;

[0037] Determine target seismic attributes by combining the target layer's planar sedimentary facies map and reservoir knowledge;

[0038] RGB color fusion is performed on the target seismic attributes to obtain target fused attribute stratum slices.

[0039] Technical effects and advantages of the present invention:

[0040] The present invention provides a method for obtaining a seismic phase map, which includes: obtaining a target fusion attribute stratigraphic slice; establishing an expert experience recognition label based on the target fusion attribute stratigraphic slice; and obtaining a seismic phase map based on the expert experience recognition label and a deep convolutional neural network deep learning technology with a U-NET structure.

[0041] This method uses deep learning technology to characterize seismic phases, avoiding the defect that some phase belts are difficult to identify, and at the same time provides a new idea for seismic phase characterization and identification.

[0042] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures pointed out in the description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 The figure is a flow chart of a method for obtaining a seismic phase map. DETAILED DESCRIPTION

[0044] The following will clearly and completely describe the technical solution of the present invention in conjunction with the accompanying drawings provided by the present invention. Moreover, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0045] It should be noted that the structures, proportions, sizes, etc. illustrated in the drawings of this specification are only used to match the contents disclosed in the specification for understanding and reading by those familiar with this technology, and are not used to limit the conditions for implementation of the present invention. Therefore, they have no substantial technical significance. Any modification of the structure, change in the proportional relationship, or adjustment of the size should still fall within the scope of the technical content disclosed by the present invention without affecting the efficacy and purpose that can be achieved by the present invention. At the same time, terms such as "upper", "lower", "left", "right", "middle" and "one" quoted in this specification are only for the convenience of description and are not used to limit the scope of implementation of the present invention. Changes or adjustments in their relative relationships should also be regarded as the scope of implementation of the present invention without substantially changing the technical content.

[0046] To address the deficiencies of the prior art, the present invention discloses a method for obtaining a seismic phase map, the method comprising: obtaining a target fusion attribute stratigraphic slice; establishing an expert experience recognition label based on the target fusion attribute stratigraphic slice; and obtaining a seismic phase map based on the expert experience recognition label and a deep convolutional neural network deep learning technology with a U-NET structure.

[0047] This method was applied to seismic facies characterization of the target layer in the shallow-water confined platform environment of the study area, effectively identifying the beach core, beach flank, tidal channel, and lagoon facies. This solved the previous problem of difficulty in distinguishing beach cores and tidal channels, and the overlap of some beach cores and beach flanks. The research results provide technical conditions for subsequent geological modeling, especially facies modeling.

[0048] To better understand this approach, let's first briefly describe its design principles: Addressing the shortcomings of conventional pattern recognition for seismic facies identification, this approach employs human-computer interactive machine learning techniques to establish a correlation between the experience of geological experts and seismic attributes, enabling seismic facies learning and characterization. First, stratigraphic slices with different attributes are extracted. Based on the planar sedimentary facies map of the target layer, seismic attributes with high correlation are selected. Different facies zones are interpreted on these attribute slices, establishing expert experience labels. Deep convolutional neural networks (CNNs) based on the U-NET architecture are then used for deep learning to obtain the final data representation of the expert model. Finally, a seismic facies map is generated through human-computer interaction.

[0049] The following combination Figure 1A method for obtaining a seismic phase map is explained in detail. The method comprises the following steps:

[0050] 1. Obtain target fusion attribute stratum slice.

[0051] Specifically: obtain seismic data volume, and then calculate amplitude, frequency, phase, coherence and other attribute volumes based on the seismic data volume; extract stratigraphic slices with different attributes based on the top and bottom interpretation horizons of the target layer; combine the planar sedimentary facies map of the target layer, select three seismic attributes with good correlation with the sedimentary facies, perform RGB color fusion, and obtain the target fused attribute stratigraphic slice.

[0052] 2. Based on the target fusion attribute stratum slice, establish expert experience recognition label.

[0053] Specifically: On the target fusion attribute slice, combined with the target layer plane sedimentary facies map and attribute plane characteristics, manual preliminary interpretation of different phase zones (referring to different microfacies, such as marine reservoir, bioclastic beach facies, beach wing facies, lagoon facies) is carried out, and expert experience recognition labels are established.

[0054] 3. Obtain seismic phase maps based on the expert experience recognition labels and deep convolutional neural network deep learning technology of U-NET structure.

[0055] Specifically: for a certain phase belt (target phase belt), the CNN deep learning technology with U-NET structure is used to extract features from attribute slices to obtain the smallest image unit that reflects the characteristics of the original input data; based on the expert experience recognition label, the feature data labeled by the expert experience is set as the final learning goal and quality control standard, and the smallest image unit is used as the training object for iterative training to obtain the final data expression that highlights the characteristics of the target phase belt.

[0056] The self-organizing neural network (SOM) technique is used to improve the final data representation by geometric distance weighting to obtain the distribution probability of the target phase belt, with a value range of (0-1). Finally, the other phase belts are predicted one by one to obtain the probability distribution of each phase.

[0057] The facies belts are merged sequentially to obtain the seismic facies map. In the area where the facies belts overlap, a certain facies belt is prioritized for synthesis based on its geological significance and facies belt morphology.

[0058] The present invention also provides a seismic phase map acquisition device, which includes: an acquisition unit for obtaining a target fusion attribute stratigraphic slice; an establishment unit for establishing an expert experience recognition label based on the target fusion attribute stratigraphic slice; and a learning unit for obtaining a seismic phase map based on the expert experience recognition label and a deep convolutional neural network deep learning technology with a U-NET structure.

[0059] Among them, the seismic data volume is obtained, and then the amplitude, frequency, phase, coherence and other attribute volumes are calculated based on the seismic data volume; based on the top and bottom interpretation layers of the target layer, stratigraphic slices with different attributes are extracted; combined with the planar sedimentary facies map of the target layer, three seismic attributes with good correlation with the sedimentary facies are selected, and RGB color fusion is performed to obtain the target fused attribute stratigraphic slice.

[0060] Among them, on the target fusion attribute slice, combined with the target layer plane sedimentary phase map and attribute plane characteristics, manual preliminary interpretation of different phase zones (referring to different microphases, such as marine reservoirs, bioclastic beach phases, beach wing phases, and lagoon phases) is carried out to establish expert experience recognition labels.

[0061] Seismic facies are obtained using the expert experience labels and deep convolutional neural network (CNN) deep learning technology with a U-NET structure. Specifically, for a specific facies belt (target facies belt), the U-NET CNN deep learning technology is used to extract features from attribute slices, obtaining the smallest image unit that reflects the characteristics of the original input data. Based on the expert experience labels, the expert experience-labeled feature data is set as the final learning target and quality control standard, and iterative training is performed on the smallest image unit as the training object to obtain the final data representation that highlights the characteristics of the target facies belt.

[0062] The self-organizing neural network (SOM) technique is used to improve the final data representation by geometric distance weighting to obtain the distribution probability of the target phase belt, with a value range of (0-1). Finally, the other phase belts are predicted one by one to obtain the probability distribution of each phase.

[0063] The facies belts are merged sequentially to obtain the seismic facies map. In the area where the facies belts overlap, a certain facies belt is prioritized for synthesis based on its geological significance and facies belt morphology.

[0064] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for obtaining a seismic phase map, characterized in that: The method comprises: Obtain target fusion attribute stratum slice; Establish expert experience recognition labels based on the target fusion attribute stratum slice; The seismic phase map is obtained by using the deep convolutional neural network deep learning technology based on the expert experience recognition labels and the U-NET structure.

2. The method according to claim 1, characterized in that Obtain target fusion attribute stratum slices, including: Get 3D attribute body; Extracting stratigraphic slices with different attributes from the 3D attribute volume based on the target layer top interpretation layer and the target layer bottom interpretation layer; Determine multiple target seismic attributes by combining the target layer's planar sedimentary facies map and reservoir knowledge; RGB color fusion is performed on the multiple target seismic attributes to obtain target fusion attribute stratum slices.

3. The method according to claim 2, characterized in that Get 3D attributes, including: Obtain 3D seismic data volume; A 3D attribute volume is obtained according to the 3D seismic data volume.

4. The method according to claim 2, characterized in that The 3D attributes include: amplitude, frequency, phase, and coherence.

5. The method according to claim 2, characterized in that Based on the target fusion attribute stratum slice, expert experience recognition labels are established, including: According to the plane sedimentary facies map of the target layer and the plane characteristics of the attributes, different facies belts are interpreted on the target fused attribute stratigraphic slice to obtain expert experience recognition labels.

6. The method according to claim 5, characterized in that Based on the expert experience recognition labels and the deep convolutional neural network deep learning technology of the U-NET structure, a seismic phase map is obtained, including: Obtain the probability distribution of each phase based on the expert experience recognition labels and the deep convolutional neural network deep learning technology of the U-NET structure; The facies belts are merged sequentially to obtain the seismic facies map.

7. The method according to claim 6, characterized in that Based on the expert experience labels and the deep convolutional neural network deep learning technology of the U-NET structure, the probability distribution of each phase is obtained, including: For the target facies belt, the deep convolutional neural network deep learning technology of the U-NET structure is used to extract features of the target fusion attribute stratum slice to obtain the smallest image unit that reflects the characteristics of the original input data; Based on the expert experience recognition labels, the feature data labeled by the expert experience is set as the final learning goal and quality control standard, and the smallest image unit is used as the training object for iterative training to obtain the final data expression that highlights the characteristics of the target phase band; According to the final data expression of the target phase belt characteristics, the probability distribution of each phase is obtained.

8. The method according to claim 7, characterized in that According to the final data expression of the target phase belt characteristics, the probability distribution of each phase is obtained, including: Using a self-organizing neural network technique to perform geometric distance weighting processing on the final data expression, to obtain the distribution probability of the target phase belt; Repeat the test for other phase zones to obtain the probability distribution for each phase.

9. A device for obtaining a seismic phase map, characterized in that: The device comprises: An acquisition unit, used for obtaining a target fusion attribute stratum slice; An establishing unit, configured to establish an expert experience recognition label according to the target fusion attribute stratum slice; A learning unit is used to obtain a seismic phase map based on the expert experience recognition label and the deep convolutional neural network deep learning technology of the U-NET structure.

10. The device according to claim 9, characterized in that Obtain target fusion attribute stratum slices, including: Get 3D attribute body; Extracting stratigraphic slices with different attributes from the 3D attribute volume based on the target layer top interpretation layer and the target layer bottom interpretation layer; Determine target seismic attributes by combining the target layer's planar sedimentary facies map and reservoir knowledge; RGB color fusion is performed on the target seismic attributes to obtain target fused attribute stratum slices.

Images