Hybrid modeling of froth flotation superficial appearance applying dynamic textures analysis

Flotation is one of the most complex processes in mining industry. In fact, non linear characteristics, large delays, significant disturbances and important unmeasurable variables make the process very difficult for automatic control. However, experimented operators are able to take adequate control actions based on superficial froth appearance by defining several operational states. To ensure the obtaining of technical and economical benefits, it is necessary to better identify different operational states from the froth superficial appearance. In this work we propose a hybrid characterization of the froth surface applying dynamic texture techniques and texture classification. Results show that the models obtained represent accurately the froth dynamics and our detection algorithm is able to detect efficiently different froth operational states.