Primary method of quadratic programming in multivariable predictive control with constraints

Abstract

General Predictive Control (GPC) is a modern method for process control which is appropriate for many characters of processes. In this paper there is proposed possibility of optimization, which is performed in each sampling period, in the GPC algorithm. Lower time of calculations is in general important for GPC control of multivariable systems with many constraints. An improvement of a primary method of quadratic programming task is proposed in this paper. Computational time can be reduced by changes in details of the optimization method. Time reserves are analyzed in a case of a nonlinear constrained problem which is represented by the Active Set Method. The improved method is presented and results are discussed in simulations. © Springer International Publishing Switzerland 2014