Application of Clustering in the Analysis and Control of Mineral Processing Plants

The control of mineral processing plants incorporates a number of off-line and on-line actions by plant personnel to obtain satisfactory plant performance. In general, off-line actions centre around intensive data analysis, while on-line activities are either automated control actions or prompted by qualitative interpretations of quantitative and qualitative information. Data sets obtained by various data logging devices and laboratory analysis of samples in many modem plants may be very large and are often unmanageable. This paper presents a method of analysing the data sets typically found in mineral processing, based on cluster analysis. The results of this analysis leads to a more complete understanding of operations in comparison with existing analysis tools used in mineral processing.Improved process configuration, optimised control strategies and in some cases, direct use of the results in monitoring process unit operations may be achieved.