Model predictive control of semiautogenous mills (sag)

The present manuscript focuses on the development of a multivariable control based on the MPC strategy for a semiautogenous grinding (SAG) device. A previously published specific SAG model that uses a deep analysis of the internal device behavior was used for the MPC strategy development. SimulinkTM software was used for the dynamic representation and control development. The selection of controlled and manipulated variables took into account performance and functional criteria. The power draw, volumetric filling level, and a size reduction percentage were the controlled variables, while the fresh ore feed rate, fresh water feed rate, and the SAG rotation speed were the manipulated variables. The controller response showed a suitable control behavior independent of the noisy multivariable modification.