Abstract

LUCAY, Freddy A; SEPULVEDA, Felipe D  and  DELGADO, José. Application with Radial Basis Neural Netwoks and Geostatistical for Interpolation/reconstruction for Database in copper grade.. REV. MAMYM [online]. 2018, n.5, pp.44-58. ISSN 2519-5352.

Abstract To quantify a mineral deposit, it is essential to determine its quality (grade) and its quantity (tonnage), so there are two stages that must be met are the following: 1) planning and execution of a mesh of a drillhole where the elements/compounds of interest are determined, 2) creation of a database and the application of mathematical techniques to calculation of the grade/tonnage of the deposit under study. These stages present the following considerations: 1) the planning of the mesh, the number of drillholes and the elements/compounds studied depends on the economic capital of the project and is carried out according to the experience of the engineer/geologist in charge, 2) The database is often created with the most relevant data, 3) The mathematical evaluation depends on the engineer´s experience and the computational tools he has. These considerations may generate uncertainty in the quality of the final result, therefore, the aim of this research is to perform an analysis in the database to improve the amount of information (interpolation and reconstruction of data), applying Neural Networks of Radial Base (NNRB), being compared their results with the Polynomial Regression (PR) and Geostatistics (Ge). The mathematical expressions and the results of the adjustments and analyzes (advantages/disadvantages. The obtained results indicate that the use of PR, was never adequate, due to the deficiency in its estimation, while the results of the NNRB, always presented very low errors and compatible with Ge, maintaining the spatial tendency of the data repositioned.

Keywords : copper grade; Mining; Interpolation/Reconstruction; drilling,; geostatistics.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )