Unified approach of equality and inequality constraints in GPC

The effect of constraints in generalized predictive control is studied with the purpose of minimizing the classical cost function subject to equality and inequality constraints. In particular, the generalized predictive control (GPC) algorithm under terminal equality constraints is reformulated in an original way which then enables a unified approach of minimization under both equality and inequality constraints. This optimization is handled by quadratic programming, where an “active set” method (or gradient projection method) is proposed to calculate the optimal future control sequence. Simulation results of a motor drive control prove the improvement of the predictive structure when taking into account saturations on the control or output signals