Markov-equivalent continuous-time GPC design

The continuous-time generalised predictive control (CGPC) using a long-horizon cost function has superior robustness as compared to some other control strategies suitable for adaptive control. The anticipated filtering (AF) of the control error within the CGPC framework can be designed so as to yield stable control systems. The main goal of this paper is to propose an AF-CGPC control design procedure making numerical designs more efficient and control more robust. The approach is based on two main principles: mean-square low-order Markovian approximation of the plant´s step response within a properly determined time interval and a rational Markov-equivalent-restricted redesign of the model of the plant.