The possibility of predicting mineral concentrations in the flotation froth by the use of realtime acquired image data and the partial least squares (PLS) regression method is investigated. This is a straightforward application of utilising image analysis in the control and monitoring of the mineral flotation process. For several reasons this approach should also have potential as an industrial application: the price of the measurement unit is relatively inexpensive, and it will quickly supply grade estimates and also important image parameters such as speed, stability, and size of the froth bubbles. However, it will probably not have the long time accuracy of the conventional on-stream analysers. To test the methodology in practise, a reasonable amount of image data was collected together with froth samples from one of the flotation cells at the Pyhasalmi mine zinc circuit in Finland. The collected images were processed off-line to extract selected features from the images, and then the PLS method was used to construct a model to predict the zinc concentration in the froth as a function of the extracted image features.
[1]
Alexander Basilevsky,et al.
Statistical Factor Analysis and Related Methods
,
1994
.
[2]
Heikki Hyötyniemi,et al.
On characterization of pulp and froth in cells of flotation plant
,
1997
.
[3]
Silvia Serranti,et al.
Using PCA In Controller Strategy Design for a Flotation Process
,
1999
.
[4]
Aldo Cipriano,et al.
A REAL TIME VISUAL SENSOR FOR SUPERVISION OF FLOTATION CELLS
,
1998
.
[5]
E. Brigham,et al.
The fast Fourier transform and its applications
,
1988
.
[6]
Chris Aldrich,et al.
The interpretation of flotation froth surfaces by using digital image analysis and neural networks
,
1995
.