Mineral processing plants are dependent on a high number of interlinking influencing factors, which must be identified and analyzed in order to ensure proper operation. Artificial neural networks and a feature analysis method are used and discussed as additional tools for the analysis of an exemplary potash compaction circuit.

Abstract

Production key figures of mineral processing plants, often designed as circuits with recirculation of material, are subject to a high number of influencing factors. In order to set up plant operation in an optimal way, identifying factors with high significance is important. In this study, an artificial neural network is employed as an additional tool for such processing plant audits by means of feature importance analysis. The presented method is applicable independently of the specific plant design, wherever sufficient process data is available. Furthermore, specific outcomes of the analysis of an exemplary potash compaction circuit are discussed.