A hybrid intelligent optimal control method for complex flotation process

In mineral industry, flotation is used to separate utilised ore from gangue. The relationships between the technical indexes, i.e. the concentrate grade and the tailing grade, and the reagent feeding appear to have strong non-linearities and uncertainties in dynamic behaviours, which can hardly be described using any accurate mathematical model. For non-automatic and non-precision of manual control of reagent feeding, technical indexes fluctuate and are difficult to be controlled within their target ranges. This article has proposed a novel hybrid intelligent optimal control method which consists of a flotation reagent intelligent setting model, a slurry consistency controller, a supply ore mass controller, an air flow rate controller and a flotation level controller, among which the flotation reagent intelligent setting model is the key which consists of a unit reagent pre-setting model-based Rule-based reasoning (RBR), a feedback compensator and a feed forward compensator RBR, and a reagent feeding computational model. The method can automatically adjust the reagent feeding when the work-condition varies, so that the technical indexes can be controlled within their target ranges. The successful application has shown that the proposed method has practical significance and high potential of being further applied in optimal control of mineral industry.