Integrated Grinding And Flotation Control - Computational Problems In An Optimum Decision Making Model

The problems involved in an optimum decision making model for flotation circuit are, first of all, to characterize the feed material in terms of parameters which can quantitatively describe the network flotation action in terms of a feasible set of control actions. Once the characterization has been postulated, it is then necessary to devise tests relating to parameter sensitivity, in order that credibility of the model predictions can be assessed. Once a credible model is available, the optimum control set is determined by a hill climbing technique. Details will be reported as to the credibility of model predictions using the stochastic flotation model on a network and further, computational problems associated with hill climbing for a complex objective function are reviewed. Discussion is presented on use of the Rosenbrock technique with constraints and use of a modified objective function with constrained derivatives when employing Fletcher and Powell technique.