Study of process dynamics in Floatex density separator based on computational intelligent vision

This paper proposes a novel computational intelligent vision–based image processing and analysis techniques that can be used in Floatex density separator (FDS), a hindered settling classifier for recovery of quality products from minerals (iron, coal) in order to develop an optimum process control strategy suitable for manipulating complex variables characterised by relatively slow process dynamics. Image processing and analysis software, MATLAB 7.0 was used to process the high speed CCD camera image. The steps include calibration, contrast enhancement and segmentation. Image features are demonstrated and correlated with voidage, particle size distribution, bed height change and density mapping under specific feed rate, pulp density and teeter flow rate of the process. All these information are used for estimation of bed pressure, underflow density cut which will help in the development of accurate control system for reliable operation of FDS from a remote control station.