A Multi-Step Output Feedback Robust Model Predictive Control for Constrained Linear Systems with Bounded Disturbance

T0637. An off-line observer is designed to estimate the true state and also guarantee the stability of the estimation error. The on-line multi-step method optimizes a sequence of controller gains, which considers future dynamics of the estimation error and bounded disturbance. The optimized controller gains guarantee robust stability of the observer system. To reduce the on-line computational burden, the controller gains and the corresponding regions of attraction are calculated for the pre-specified estimated states and bounds of the estimation error. These controller gains and the corresponding regions of attraction are stored in a look-up table. On-line, the controller gains are searched in this look-up table according to the region of attraction with the closest containment of the real time estimated sate and bounds of the estimation error, which significantly reduces the on-line computational burden.