Discussion on: 'Robust Design of Predictive Controllers in Presence of Unmodeled Dynamics'

摘要

In the paper by Canale and Milanese the robustness analysis of a model predictive controller (MPC) algorithm is presented. The MPC algorithm uses an approximate model of the actual plant to both estimate states via an observer and/ormulate an receding finite horizon regulation control problem. Due to the use of an approximate model, a modeling error is unavoidable and the MPC algorithm is labeled robust by formulating conditions on state contraction and size of the model error to guarantee stability and feasibility of the MPC.