Pole-placement optimizing controller and application to a simulated autogeneous grinding circuit

An optimizing controller is developed in order to operate the process at its optimum according to an unconstrained criterion despite disturbances. The gradient of the criterion is taken as the controlled variable and a null set point is used in order to maximize the criterion. A local dynamic moving average model is identified and its gain is calculated to evaluate the gradient on-line. A low-forgetting factor in conjunction with a fixed low-pass filter is used in order to fix the dynamic response of the measured gradient. A multivariable controller using a pole cancellation method is used to achieve zero gradient set point. The method is applied to the optimization of the power drawn by a simulated autogeneous grinding mill.