Hybrid intelligent optimal control for flotation processes

In flotation process, the concentrate grade and the tailing grade are crucial technical indices and reflect the product quality and efficiency. The technical indices hardly be measured online continuously varying with the process variables and boundary conditions. Moreover, there are strong nonlinearity and uncertainty between such technical indices and the process variables, which are difficult to be described by accurate mathematical model. Therefore conventional control methods are incapable of keeping the actual technical indices within their target ranges. To solve this problem, a hybrid intelligent optimal control method is presented for flotation process. This method consists of four modules, namely a pre-setting model based on CBR (case-based reasoning), a feedback compensation model based on RBR (rule-based reasoning), a feedforward compensation model based on RBR and a soft sensor with RBF (radial basis function) neural network. The proposed approach has been successfully applied to flotation process in a hematite ore processing plant in China, and its effectiveness has been proved evidently.