Modeling and Prediction of Iron Ore Quality Indicators

Abstract

The paper proposes solution of the topical scientific problem that consists in developing a geometrical method of predicting quality indicators of iron ore deposits, applying a mathematical model of a multidimensional random geochemical field which is realized on the basis of self-organizing prediction methods. The authors develop a multidimensional heuristic prediction algorithm that uses a polynomial of arbitrary power and enables description of any functional dependency. It is demonstrated that a system of equations of a multidimensional random geochemical field should be used to mathematically describe elements of the rock massif. The grapho-analytical model of the deposit is built using geostatistical methods. It is determined that at Kryvbas deposits the kriging method is the most suitable for assessing and improving reliability of the input geological data since detailed geological exploration is carried out by means of an irregular grid of boreholes. An important aspect of geometrization of iron ore deposits is geometrical prediction of their quality indicators for solving tasks of long-term and current planning in order to provide the most efficient performance of the mining enterprise to improve rationalization of deposit development.