EHMAC -- a new simple tool for robust linear multivariable control (Q2703064)

A new predictive control strategy, the extended horizon model algorithmic control (EHMAC) is presented. EHMAC is designed to control robust open loop stable nonminimum-phase (NMP) multivariable systems under very large model-plant mismatches. EHMAC is presented as an effective combination of a long-range predictive control (EHPC) and internal model control (MAC). EHMAC is designed at two separate stages. The first stage is coming from EHPC and is aimed at stabilization of a closed-loop system incorporating a NMP model. The second stage selects a filter as suggests MAC and is aimed at providing control robustness even for extremely large deviations. The selection of a prediction horizon and a filter parameter are made straightforward by suggested procedures of auto-tuning. The numerical simplicity and other good features of the EHMAC algorithm are illustrated by examples.