Recognition of cluster's optimal quantity in case of gustafson-kessel's method usage to define technological variety of iron ore

Abstract

It was considered the problem of cluster analysis in condition of equivocation to forming automatic control process sorting large-size iron ore raw material with operating recognition of its separate pieces technological variety in stream on conveying belt. The most preferred algorithms are algorithms, which set quantity of clusters by themselves, or self-organization algorithms, one of which is Gustafson-Kessel’s algorithm. Evaluation test of clusterization was performed with using scalar measures of reliability. It was carried out the experimental researches of quality performance from cluster’s quantity for definitions technological variety of iron ore. According to our research the result was established, that the optimal clusters quantity, during identifying technological variety of iron ore raw materials, ore’s hematite variety in occurrence of Krivyi Rih, is 7.