An implementation of generalized predictive control in a flotation plant

A single-input/single-output (SISO) GPC algorithm and an ARX model identifi cation routine were programmed in the form of a portable "module" on a grap hical computing environment. Although attractive from an end-user (operator ) standpoint, the selected platforms presented some programming challenges. The whole GPC module (controller and identification) was implemented on an industrial control system and applied to many sulfide flotation circuits i n the Brunswick Mining concentrator (New Brunswick: Canada). The performanc e of the GPC controller was good on all the flotation circuits to which it was applied. Significant reductions of grade variance are reported and the controller exhibits good load and servo response. (C) 2000 Elsevier Science Ltd. All rights reserved.