CN-122017970-A - Sea sand automatic identification method based on multi-source geophysical data fusion

CN122017970ACN 122017970 ACN122017970 ACN 122017970ACN-122017970-A

Abstract

The invention relates to the technical field of automatic sea sand identification, in particular to an automatic sea sand identification method based on multi-source geophysical data fusion. The method comprises the following steps of obtaining drilling data, single-channel seismic data and shallow stratum profile data of a target area, preprocessing to generate a fusion data body, constructing a two-channel deep learning model based on the fusion data body, training the two-channel deep learning model by using lithology tags calibrated in the fusion data body and the drilling data, identifying the target area by using the trained two-channel deep learning model, verifying an identification result, and generating sea sand distribution results.

Inventors

CHEN CHENG
WANG YING
ZHAO KUN
JIA LIYUN
ZHANG HAO

Assignees

中国地质科学院地质力学研究所

Dates

Publication Date: 20260512
Application Date: 20260126

Claims (10)

1. A sea sand automatic identification method based on multi-source geophysical data fusion is characterized by comprising the following steps: s1, acquiring drilling data, single-channel seismic data and shallow stratum profile data of a target area, and preprocessing to generate a fusion data body; s2, constructing a two-channel deep learning model based on the fusion data body; s3, training a two-channel deep learning model by using lithology tags calibrated in the fusion data body and the drilling data; and S4, identifying the target area by using the trained two-channel deep learning model, and verifying the identification result to generate sea sand distribution results.
2. The automatic sea sand identification method based on multi-source geophysical data fusion according to claim 1, wherein step S1 comprises the steps of: S11, acquiring drilling data, single-channel seismic data and shallow stratum section data of a target area, and preprocessing; s12, analyzing single-channel seismic data and shallow stratum profile data by taking drilling data as a reference, and establishing a corresponding relation between lithology and geophysical response; Step S13, based on the corresponding relation, single-channel seismic data and shallow stratum profile data are projected to the same geographic coordinate system, and space registration data are generated; and S14, converting the single-channel seismic data in the spatial registration data from a time domain to a depth domain to generate a fusion data volume.
3. The automatic sea sand identification method based on multi-source geophysical data fusion according to claim 2, wherein step S12 comprises the steps of: step S121, extracting space position information of sand layers and mudstones from single-channel seismic data and shallow stratum profile data based on drilling data; step S122, based on the space position information, the reflection characteristic parameters of the single-channel seismic data and the penetrability characteristic parameters of the shallow stratum profile data are counted, and the corresponding relation between lithology and geophysical response is established.
4. The automatic sea sand identification method based on multi-source geophysical data fusion according to claim 2, wherein step S13 comprises the steps of: s131, obtaining geographic reference information of single-channel seismic data and shallow stratum profile data in the corresponding relation; And S132, respectively performing coordinate transformation on the single-channel seismic data and the shallow stratum profile data based on the geographic reference information, and generating spatial registration data by using uniform geographic coordinate references.
5. The automatic sea sand identification method based on multi-source geophysical data fusion according to claim 2, wherein step S14 comprises the steps of: step S141, acquiring well logging speed information corresponding to the spatial registration data; Step S142, carrying out time-depth conversion processing on single-channel seismic time data in the spatial registration data based on the well logging speed information to generate depth domain seismic data; And S143, merging the depth domain seismic data with the shallow stratum section data in the spatial registration data to generate a fusion data volume.
6. The automatic sea sand identification method based on multi-source geophysical data fusion according to claim 1, wherein step S2 comprises the steps of: s21, constructing a network channel of the earthquake time sequence characteristic based on the characteristic of single-channel earthquake data in the fusion data body; s22, constructing a network channel of image profile characteristics based on the characteristics of the shallow stratum profile data in the fusion data body; and S23, constructing a fusion network layer based on the network channel of the image profile characteristics and the network channel of the earthquake time sequence characteristics to form a complete two-channel deep learning model.
7. The automatic sea sand identification method based on multi-source geophysical data fusion according to claim 6, wherein step S21 comprises the steps of: step S211, carrying out format check and data integrity verification on single-channel seismic time series data in the fusion data body; Step S212, constructing a one-dimensional convolution network structure based on the single-channel seismic time sequence data after verification, and forming a network channel of the seismic time sequence feature.
8. The automatic sea sand identification method based on multi-source geophysical data fusion of claim 6 wherein step S22 comprises the steps of: step S221, carrying out standardization processing on the shallow stratum section image data in the fusion data body to obtain standardized shallow stratum section image data; step S222, based on the format characteristics of the standardized shallow stratum section image data, constructing a two-dimensional convolution network channel comprising a convolution layer and a pooling layer.
9. The automatic sea sand identification method based on multi-source geophysical data fusion according to claim 1, wherein step S3 comprises the steps of: s31, taking lithology labels calibrated in the fusion data body and the drilling data as input samples, and performing iterative training on a two-channel deep learning model; And S32, adding a penetration depth threshold constraint in the shallow stratum profile data in the model training process, and correcting a prediction result of the model.
10. The automatic sea sand identification method based on multi-source geophysical data fusion according to claim 1, wherein step S4 comprises the steps of: s41, inputting the multi-source data processed by the target area into a training double-channel deep learning model to obtain preliminary sea sand identification data; And S42, comparing and analyzing the preliminary sea sand identification data with the verification drilling data to generate sea sand distribution results.

Description

Sea sand automatic identification method based on multi-source geophysical data fusion Technical Field The invention relates to the field of marine geological exploration, in particular to a sea sand automatic identification method based on multi-source geophysical data fusion. Background Sea sand is used as an important marine mineral resource, is widely applied to the fields of construction, road construction and the like, and has important significance for reasonable development and utilization of resources through accurate identification and distribution investigation. The traditional sea sand identification method mainly relies on manual analysis of seismic data and drilling data, and has the problems of low efficiency, strong subjectivity, large influence on identification accuracy by experience and the like. In the prior art, although single seismic data are combined with machine learning to carry out lithology recognition, single data type has the limitations that single seismic data are easy to be interfered by noise, the recognition precision of a thin sand layer is insufficient, shallow stratum section data are mostly used for qualitative analysis and are not effectively fused with data, quantitative utilization of physical characteristics of 'mud penetration and sand penetration prevention' of the shallow stratum section is lacked, mud rock is not completely removed, sea sand recognition misjudgment rate is high, the effectiveness of model training is limited by sparsity of drilling data, and the requirement of accurate sea sand exploration in a large area is difficult to meet. Therefore, how to fully integrate complementarity of multi-source geophysical data, realize high-efficiency and accurate automatic identification of sea sand by using a deep learning technology, strengthen the mud rock removal effect by means of penetrability difference of shallow stratum sections, and become a technical problem to be solved in the current marine geological exploration field. Disclosure of Invention Based on this, it is necessary to provide an automatic sea sand identification method based on multi-source geophysical data fusion, so as to solve at least one of the above technical problems. In order to achieve the above purpose, the sea sand automatic identification method based on multi-source geophysical data fusion comprises the following steps: s1, acquiring drilling data, single-channel seismic data and shallow stratum profile data of a target area, and preprocessing to generate a fusion data body; s2, constructing a two-channel deep learning model based on the fusion data body; s3, training a two-channel deep learning model by using lithology tags calibrated in the fusion data body and the drilling data; and S4, identifying the target area by using the trained two-channel deep learning model, and verifying the identification result to generate sea sand distribution results. The method has the beneficial effects that the sea sand resource identification precision and efficiency are obviously improved through the deep fusion and intelligent analysis of the multi-source geophysical data. Firstly, a fusion data body with an accurate spatial mapping relation is constructed by establishing a spatial registration and scale normalization mechanism of drilling data, single-channel seismic data and shallow stratum profile data, so that the technical problems of non-uniform spatial reference and large scale difference of multi-source data in the traditional method are solved, and a reliable data base is provided for subsequent analysis. Secondly, a dual-channel deep learning model architecture is innovatively designed, seismic time sequence features and shallow-profile image features are optimized and extracted respectively, the effective integration of multi-dimensional information is realized through a feature fusion layer, the defect that the feature expression capability of a single data source is limited is overcome, and the recognition capability of the model to complex geological features is remarkably enhanced. Drawings FIG. 1 is a schematic flow chart of steps of a method for automatically identifying sea sand based on multi-source geophysical data fusion; FIG. 2 is a flowchart illustrating the detailed implementation of step S2 in FIG. 1; FIG. 3 is a flow chart for automatic identification of sea sand; The achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments. Detailed Description The following is a clear and complete description of the technical method of the present invention, taken in conjunction with the accompanying drawings, and it is evident that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without