Generalized Predictive Control of Multivariable Systems

摘要

The multivariable extension of the generalized predictive control strategy of Clarke et al., (1984) is considered. It is formally shown that, subject to a suitable choice of (finite) output and control horizons, a quadratic cost function is minimized without any prior knowledge of the system delay matrix. Simulation results on two multivariable systems demonstrate the validity of the method.