Super voxel digital outcrop configuration clustering analysis method based on point cloud optimization
Through the point cloud-optimized super voxel digital outcrop configuration cluster analysis method, the problems of data accuracy and field geological complexity of three-dimensional laser scanners are solved, and high-resolution recognition of outcrop and rapid deposition of microphase types are achieved. Quantitative description improves the efficiency and accuracy of oil and gas exploration.
Patent Information
- Application Number
- CN202410896372.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-05
- Publication Date
- 2024-11-12
AI Technical Summary
The point cloud data obtained by the three-dimensional laser scanner has accuracy problems in data processing, and the deposition complexity and configuration of field geological outcrops are irregular, making it difficult to quickly and quantitatively describe the size, morphology and the size and morphology of different deposition microphase types. Distribution affects oil and gas exploration and development.
Using the super voxel digital outcrop configuration clustering analysis method based on point cloud optimization, data is collected through a three-dimensional laser scanner, point cloud distance regularization, downsampling, Kriging interpolation, multi-attribute feature fusion and super voxel feature Dimension reduction, clustering parameters are determined, configuration clustering analysis is performed to identify lithologies and sedimentary microfacies types of outcrop.
High-resolution identification of outcrop profiles is achieved, the size and distribution of deposited microfacial types is rapidly and quantitatively described, the efficiency and accuracy of oil and gas exploration is improved, sandstone reservoirs can be effectively identified, and oil and gas development is supported.
Smart Images

Figure CN118941828A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of oil and gas exploration, and is a method for identifying the lithology of an outcrop profile point cloud by using an optimized supervoxel digital outcrop configuration clustering analysis method, thereby dividing the morphological size and distribution of the sedimentary microfacies types of the outcrop profile. Background Art
[0002] The 3D laser scanner emits laser light onto the surface of the outcrop and obtains information through the reflected laser signal. The laser radar scanning can accurately calculate the spatial position of each point on the outcrop surface and collect tens of thousands of points to generate a point cloud that is close to the shape of the target surface. The position of any point of the returned laser light can be recorded as (x, y, z) coordinates, reflection intensity (I) and color (R, G, B), etc. Compared with traditional field data measurement technology, the laser radar has real-time, precise and structured perception capabilities. The advantages are: (1) The collected data is completely collected for the three-dimensional space where the outcrop is located, and no specific points or specific scanning lines are set to collect data points; (2) Data can be collected in high altitude areas that are difficult to survey; (3) The field conditions at the time are objectively recorded, which can be used for later re-survey and re-interpretation; (4) Data can be collected quickly. Point clouds are collected from real field outcrops, which naturally contain the structural priors of the real scene.
[0003] In addition to providing detailed spatial data for 3D modeling, the spectral capability of LiDAR can also be considered. At the same spectral wavelength, the intensity echo (reflected power / transmitted power) of field outcrops is mainly affected by the target distance and reflectivity, which is specifically composed of the composition of the sedimentary rock surface particles and the size of the particle size of the outcrop profile. When the photons of the emitted light source encounter the outcrop profile, some photons are absorbed and some photons are scattered. The larger the particle size, the more scattered photons and the greater the reflection intensity; the smaller the particle size, the fewer scattered photons and the smaller the reflection intensity. So in theory, sandstone is more reflective than mudstone and shale. Therefore, the LiDAR intensity of the point cloud data of the 3D laser scanner can be regarded as a lithology identification technology that is worth studying. The main research area is the Yanchang Formation in the southern part of the Ordos Basin.
[0004] The outcrop profile can intuitively and completely display its sedimentary characteristics, lay a good foundation for further fine characterization of the internal structure of the reservoir, establish a sand body configuration model, and have a good demonstration effect on the study of underground sand body configuration. Among them, the sand body configuration is an effective scheme to solve the matching of a certain sedimentary age in the sedimentary stratum, the sedimentary microfacies and the cumulative thickness of the reservoir. Sedimentary microfacies characteristics are the main content of oil and gas development. Favorable sedimentary microfacies have a significant impact on the development of reservoirs. The sedimentary microfacies type of high-quality sandstone reservoirs is generally an underwater distributary channel. Different sedimentary paleogeographic environments (sedimentary microfacies) of outcrops bring different sediments. There are differences in the point cloud features of the 3D laser scanner, that is, there are differences in the lithological characteristics of the sediments; according to the indication of the lithological characteristics, the sedimentary environment is analyzed, the sedimentary microfacies type is determined, and the outcrop configuration model is divided. In general, the different regional characteristics of the point cloud digital outcrop represent different configurations of the actual outcrop profile. However, the point cloud data obtained by the 3D laser scanner still has accuracy issues, which brings challenges to data processing. In addition, field geological outcrops also have sedimentary complexity and irregular configuration morphology. Therefore, through optimized method processes, high-resolution point cloud digital outcrops are constructed to perform lithology identification, and quickly and quantitatively describe the size, morphology and distribution of different sedimentary microfacies types, thereby identifying sandstone reservoirs in sedimentary outcrops and achieving efficient oil and gas exploration and development. Summary of the invention
[0005] The technical problem to be solved by the present invention is to propose a supervoxel digital outcrop configuration clustering analysis method based on point cloud optimization.
[0006] 1. In order to achieve the above purpose, the following technical scheme is adopted, and the specific steps are as follows:
[0007] Step 1: Use a 3D laser scanner to collect data on the outcrop profile in the field, obtain spatial position geometric information, and reconstruct the outcrop point cloud surface model.
[0008] Step 2: Regularize the point cloud distance of the reconstructed digital outcrop model.
[0009] Step 3: Set the radius and threshold for downsampling the point cloud number of the reconstructed digital outcrop model.
[0010] Step 4: Perform Kriging interpolation on the outcrop point cloud, which can be expressed by the following relationship:
[0011]
[0012] Where P represents an unknown point, and its value Z(P) can be obtained by the known point λ i Value Z(P i ) and the spatial correlation between them to estimate, λ iis the weight coefficient determined based on the semivariogram.
[0013] Step 5: Multi-attribute feature fusion is performed on different feature values of the point cloud according to different weights to generate point cloud supervoxel outcrops.
[0014] Step 6: Reduce the dimension of supervoxel features and determine clustering parameters.
[0015] Step 7: Select appropriate outcrops for configuration cluster analysis training and verification to see whether they are close to the actual outcrop profile results.
[0016] Step 8: According to the training results, determine the radius, threshold and weighting parameters in the clustering method and output them.
[0017] Step 9: Perform a cluster analysis of the configuration of the remaining generated point cloud supervoxel digital outcrops. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] Figure 1 Flowchart of the supervoxel digital outcrop configuration clustering analysis method for optimizing point clouds;
[0019] Figure 2 Sedimentary microfacies division map of the 8th phase outcrop of Shiwanghe River in the Yichuan section of the Ordos Basin;
[0020] Figure 3 Reconstruction of point cloud surface model of the 8th phase outcrop of Shiwang River in Yichuan section of Ordos Basin;
[0021] Figure 4 The result of cluster analysis of the outcrop configuration of the Shiwang River in the 8th phase of the Yichuan section in the Ordos Basin;
[0022] Figure 5 The extended section of the Ordos Basin, Erlipan, is 8 meters long. 1 Period outcrop field survey;
[0023] Figure 6 The extended section of the Ordos Basin, Erlipan, is 8 meters long. 1 Reconstruction of surface model of outcrop point cloud;
[0024] Figure 7 The extended section of the Ordos Basin, Erlipan, is 8 meters long. 1 Results of cluster analysis of outcrop configuration of point cloud supervoxel;
[0025] Figure 8 Field survey of the 81st outcrop of the Hukou service area in the Yichuan section of the Ordos Basin;
[0026] Fig. 9 Reconstruction of the point cloud surface model of the outcrop of the 81st phase of the Hukou service area in the Yichuan section of the Ordos Basin;
[0027] Fig.10 The result of cluster analysis of the outcrop configuration of the point cloud supervoxel in the Hukou service area of the Yichuan section in the Ordos Basin for the 81st period; DETAILED DESCRIPTION
[0028] Step 1: Field survey, profile data collection, use a 3D laser scanner to obtain the point cloud information of the target outcrop profile, and the position of the scanning station needs to be reasonably set according to the length and width of the field outcrop. In order to improve the accuracy of scanning the outcrop, several stations are set for scanning. The scanned part of each station has a 30%-50% overlap. Data splicing is performed, and several points with similar distances are selected to ensure full coverage of the outcrop. In actual field operations, noise points and outliers are inevitably present due to various factors. For example, vegetation on the outcrop profile and houses and buildings next to it. These noise points are relatively obvious and isolated, easy to identify, and are directly processed by manual elimination to obtain the exposed stratigraphic rock point cloud. Then reconstruct the point cloud surface model of the outcrop and input it. The outcrops surveyed in the field all belong to the Triassic Yanchang Formation in the Ordos Basin.
[0029] Step 2: Regularize the distances of the reconstructed outcrop point cloud model at different heights and with different surface slopes, thereby repairing local classification errors in the point cloud supervoxels.
[0030] Step 3: Due to the excessive amount of point cloud data, in order to reduce errors and improve the accuracy of configuration cluster analysis, the main function is to achieve dimensionality reduction of feature information by compressing the number of parameters, which can effectively reduce overfitting and thus improve the fault tolerance. Since the outcrop sections of the Yanchang Formation surveyed in the field present different geometric features (such as concavity and convexity), different downsampling methods should be adopted according to the heterogeneity of different outcrops to be close to the actual outcrop geological phenomena.
[0031] Step 4. The above sampling methods are uniform sampling methods based on spatial geometric position characteristics. In the process of downsampling, the effective point cloud data of the digital outcrop profile configuration may be reduced. Therefore, in order to make the digital outcrop characteristics more obvious, that is, to increase the amount of effective point cloud data, it is necessary to perform spatial attribute interpolation on the downsampled point cloud. The Kriging spatial interpolation method is used to estimate the attribute values of unknown points based on the attributes of known points and use spatial correlation to generate a continuous surface model.
[0032] Step 5: The different eigenvalues of the point cloud are fused with multi-attribute features according to different weights to generate point cloud supervoxel outcrops. In order to increase the configuration clustering effect, the supervoxel eigenvalues are normalized by dimensionality reduction. The weighting parameters are first set empirically, and then continuously adjusted according to the experimental results during the clustering analysis until the appropriate weighting parameters are selected. At the same time, the appropriate number of clusters is set to control the effective granularity and accuracy of the clustering. The clustering types are set as sandstone, siltstone and mudstone.
[0033] Step 6: Select outcrops for configuration cluster analysis training and verification. The outcrops selected for training are the 8th phase outcrops of Shiwang River in Yichuan section. The lithology and configuration mode of the outcrops have been determined in the field survey. The parameter settings of the above steps are all tested on this outcrop, and the radius R, threshold and weighting parameters are finally determined. The experimental results show that the deposits in the upper and middle parts of the outcrop are fine sandstone, and the deposits in the lower part are siltstone; the sedimentary microfacies types are estuary bar-underwater diversion channel from bottom to top, and the experimental results are close to the actual survey results.
[0034] Step 7: Output the training results and perform cluster analysis on the configuration of the remaining Yanchang Group point cloud supervoxel digital outcrops. The geographical locations of the two outcrops analyzed are located near Erlipan in Yanchang County and Hukou Service Area in Yichuan County. Finally, the configuration division of the two field outcrops is completed to clarify the type and size of the sedimentary microfacies. 1 The outcrops of the period show obvious lithological divisions of sand and mud. The thickness of sandstone increases significantly from bottom to top, and the entire outcrop section shows reverse cyclic characteristics. Massive sandstone is developed in the middle and upper part, sandy mudstone and weathered deposits are developed in the lower left part, and sandstone and mudstone are filled in the lower right part. Overall, the sedimentary microfacies types of the outcrops are underwater distributary channels and distributary bays from top to bottom, which is similar to the results of the field survey of the outcrop section. 1 Thick layers of fine sandstone are developed in the outcrop section, and the sedimentary microfacies type is multi-stage underwater distributary channel.
Claims
1. A method for modeling sand body configuration of aerial outcrops based on improved multilinear interpolation method, characterized in that Use the following steps: (1) Use a 3D laser scanner to collect data on the outcrop profile in the field, obtain spatial position geometric information, and reconstruct the outcrop point cloud surface model. (2) Regularization of point cloud distance of reconstructed digital outcrop model (3) Set the radius and threshold for downsampling the point cloud of the reconstructed digital outcrop model (4) Kriging interpolation of point cloud can be expressed by the following relationship: Where P represents an unknown point, and its value Z(P) can be obtained by the known point λ i Value Z(P i ) and the spatial correlation between them to estimate, λ i is the weight coefficient determined based on the semivariogram. (5) The different feature values of the point cloud are fused according to different weights to generate the point cloud super-voxel outcrop. (6) Reduce the dimension of supervoxel features and determine clustering parameters. (7) Select outcrops for configuration cluster analysis training and verification to see whether they are close to the actual outcrop profile results. (8) According to the training results, the radius, threshold and weighting parameters in the clustering method are determined and output. (9) Perform a cluster analysis of the configuration of the remaining generated point cloud supervoxel digital outcrops.
