Font Size: a A A

Construction Method For Three-dimensional Rock Formation Modeling Based On Multi-source Data

Posted on:2023-05-19Degree:MasterType:Thesis
Country:ChinaCandidate:Z W YiFull Text:PDF
GTID:2530307070487564Subject:Engineering
Abstract/Summary:PDF Full Text Request
The three-dimensional model can directly and accurately reflect the geological structure and boundary conditions of underground strata,which plays an important role in the exploration of geological work.In view of the use of limited geological exploration data are difficult to build 3D model accord with the actual rock strata,based on the drilling data as the main data,gravity and magnetic data such as geological profile,as the auxiliary,combined with machine learning algorithms,reveal the geological characteristics of multi-source data to forecast the virtual borehole lithological distribution,and use the radial basis 3D model surface reconstruction method to establish rock strata,The method of constructing 3D model of rock strata based on multi-source data is studied.The main research work and achievements of this paper are as follows:(1)On the basis of digitization of geological data and resampling of borehole data in the study area,the strata interface points are extracted;The 3D data layer is formed by upward continuation of gravity and magnetic data,and the sample data set for 3D rock layer modeling assisted by gravity and magnetic data is constructed.(2)Aiming at the problem of unbalanced sampling distribution of borehole data,the over-sampling method for unbalanced data processing and the machine learning algorithm for classification and recognition of rock strata attributes are compared and studied.Adasyn-catboost method is used to build the rock strata modeling process based on borehole data.On the basis of borehole data,geological profile and gravity and magnetic data are supplemented,and combined with CGAN deep learning method,strata attribute discrimination and 3D modeling process assisted by gravity and magnetic data are studied.(3)Combined with the exploration data and gravity and magnetic data of zhaoxian gold deposit,virtual drilling was introduced,and the above machine learning method was adopted to realize the identification of rock strata attributes,and a 3D model of rock strata was constructed based on hidden function surface.Through profile calculation model and the actual contact ratio between prospecting line section,analysis of the rock based on geological exploration data,only three dimensional implicit model and based on multi-source data support to build three-dimensional model of strata,the results show that the proposed in this paper based on multi-source data constraints of rock three-dimensional model building method can effectively improve the reliability and precision of the strata of 3D modeling.The above research results provide a new idea for 3D geological modeling with multi-source data fusion.
Keywords/Search Tags:Multi-Source Data, Three-dimensional stratum model, Drilling data, Machine learning, CGAN
PDF Full Text Request
Related items