Controlling a grinding mill circuit using constrained model predictive static programming

Abstract

A constrained Model Predictive Static Programming (MPSP) method is implemented in simulation to a single-stage grinding mill circuit model. The results are compared to a constrained Nonlinear Model Predictive Control (NMPC) method. Both the constrained MPSP and NMPC controllers were able to track the desired output set-points without exceeding any constraints. The comparison shows that the constrained MPSP has a faster computational time than that of the NMPC controller with similar performance. Therefore, constrained MPSP shows promise as a model-based controller for large processes where computational time limits the use of NMPC.