Retrieving Information from Time Dependent Mineral Processing Data.

The new generation of computer environment provides mineral processing plants with convenient tools to store and retrieve or to analyze on-line process variable values as given along time by the various sensors installed in the plant. The paper describes the application of classical tools, such as time series analysis, to retrieve the hidden information contained is such data. The concepts used in single signal processing (autocorrelogram, power spectrum, ARIMA descriptors) and multi-signal processing (cross-correlogram, cross-spectrum) are briefly reviewed. Some examples are given to illustrate their application (sampling, data filtering, process behaviour prediction, process evaluation, data synchronisation). Examples are taken from mixing, flotation and leaching processes.