Прогноз и оценка рудоносности при помощи непараметрического регрессионного анализа, оценка векторного потенциала и его градиента

S. Bakurdziev, L. Ratiev, О. Zakhariev, D. Vundev, P. Mateev. Prognostication and estimation of ore-bearing by means of nonparametric regression analysis, estimation of vector potential and its gradient. Two methods- kriging and trend-analysis are broadly used in ore-bearing estimation and prognostication. Both have their advantages and disadvantages. A methodology combining advantages of above methods and minimizing their shortcomings is proposed and discussed here. For the purpose a program NPLRG is written (by Vundev and Mateev) providing: 1) prognostication within three-dimensional regular (or irregular) grid; 2) field selection; 3) scaling along X, Y, Z; 4) selection of weight function; 5) selection of the number of variables (M ≤ 1O) and of their combinations. At any given point are found: 1) number of observations evaluated in calculations; 2) prognostication of the mean content; 3) 5-% level of significance of the prognostication; 4) prognostication gradient along three axes and its variance. This method was tested on ore bodies of different type – stratiform, lenses, veins, porphyry deposits, ore-bearing faults, ore columns, deep marine deposits of manganese concretions. The program provides also simulation of ore bodies of any form and display of numeric and graphic data using microprocessor technics.